Racial and ethnic disparities in alcohol-associated liver disease in the United States: A systematic review and meta-analysis

Abstract

Background:

Alcohol-associated liver disease (ALD), encompassing alcohol-associated hepatitis and alcohol-associated cirrhosis, is rising in the United States. Racial and ethnic disparities are evident within ALD; however, the precise nature of these disparities is poorly defined.

Methods:

We conducted a search of the PubMed/MEDLINE and EMBASE databases to identify studies published from inception through September 2023 that reported ALD incidence, prevalence, and mortality within the United States, stratified by race and ethnicity. We calculated pooled prevalence and incidence by race and ethnicity, including risk ratios and ORs for ALD pooled prevalence and alcohol-associated hepatitis/alcohol-associated cirrhosis pooled proportions, and OR for ALD mortality using the DerSimonian and Laird method for random-effect models.

Results:

We identified 25 relevant studies (16 for quantitative meta-analysis), comprising 76,867,544 patients. ALD prevalence was highest in Hispanic (4.5%), followed by White (3.1%) and Black (1.4%) individuals. Pooled risk ratios of ALD prevalence were 1.64 (95% CI: 1.12–2.39) for Hispanic and 0.59 (95% CI: 0.35–0.87) for Black compared to White individuals. Mortality among those with ALD did not significantly differ between White and Hispanic (OR: 1.54, 95% CI: 0.9–2.5; I ²=0%), Black (OR: 1.2, 95% CI: 0.8–1.6; I ²=0%), or Native American (OR: 2.41, 95% CI: 0.9–2.9) individuals, while there was a significant difference between White and Asian (OR: 0.1; 95% CI: 0.03–0.5) individuals. Most data were cross-sectional and assessed to be of poor or fair quality.

Conclusions:

Differences were observed in ALD epidemiology, including higher prevalence among Hispanic and lower prevalence among Black individuals, although there were smaller differences in ALD mortality. Differences in ALD prevalence and prognosis remain poorly defined based on existing data, highlighting a need for higher-quality epidemiological studies in this area.

INTRODUCTION

Alcohol-associated liver disease (ALD) continues to increase in the United States and has become the number one cause of death from cirrhosis and the leading indication for liver transplantation.¹,²,³ Notably, ALD prevalence among patients with advanced fibrosis (stage 3 or greater) increased from 2.2% to 6.6% from 2001 to 2016.⁴ Similarly, the US prevalence of alcohol use disorder (AUD), a leading cause of ALD, has also increased and in 2018 was estimated at 5% among individuals over the age of 12 years.⁵ ALD is a spectrum ranging from early disease stages (ie, alcohol-associated fatty liver [AFL]) to later stages (ie, alcohol-associated hepatitis [AH] and alcohol-associated cirrhosis [AC]). Heavy alcohol consumption (typically defined as >4 standard drinks/d for women and >5 standard drinks/d for men) results in AFL, which is reversible with alcohol abstinence and has a favorable prognosis.⁶ However, with ongoing alcohol exposure, individuals with fatty liver disease (steatosis) can progress to AH (10%–35%) or AC (8%–20%), which carries significant morbidity and mortality risk.⁷,⁸ The prevalence of early stages of ALD in the United States has been estimated to be ~8%.⁴ However, a major setback to the proper assessment of patients with ALD has been the difficulty in identifying earlier stages (both clinically and diagnostically) of patients with ALD, when interventions can be most impactful.⁹,¹⁰

Despite the increasing concern for racial and ethnic disparities among patients with ALD in the United States, the field remains understudied. For example, ALD appears to disproportionately affect American Indian (AI) individuals in terms of incidence, severity, and morbidity, as well as Hispanic individuals in terms of incidence, morbidity, and possibly severity (defined by the presence of AH or AC).¹¹,¹²,¹³,¹⁴ Specifically, AIs have been shown to have a 5-fold increase in ALD mortality compared with non-Hispanic White (NHW) patients (49 deaths/100,000 vs. 9 deaths/100,000, respectively),¹⁵ while Hispanic populations have been shown to have a higher ALD prevalence (1.6%) compared with NHW populations (1.2%),¹³ and highest among young Hispanic individuals (prevalence 7.5%, compared to 5.4% among NHW individuals).¹⁶ There are several conflicting studies in the literature that merit consideration. First, the long-held belief that Hispanic patients have lower ALD mortality despite a higher burden of disease (the “Hispanic paradox”) has been questioned with more recent data showing increased mortality among Hispanic patients, including Hispanic men having the highest age-adjusted alcohol-associated mortality rate (41.6 per 100,000) compared to NHW populations.¹¹ Second, many studies have suggested that non-Hispanic Black (NHB) populations have reduced ALD burden (2.5% prevalence, ie, >2-fold less than other racial/ethnic groups)¹⁶ and the lowest ALD age-adjusted mortality rates (8.4 and 3.7 per 100,000 individuals among NHB men and women, respectively)¹⁷; however, other studies have demonstrated similar or increased ALD mortality among NHB individuals compared to among NHW individuals.¹⁸

Given this outcome heterogeneity in the existing studies, varying sample sizes, and population types, a comprehensive review and meta-analysis across all studies is needed to better define these racial and ethnic differences and help ascertain the extent of the underlying disparities and guide future research. Therefore, the aim of this meta-analysis was to ascertain racial and ethnic differences in ALD incidence, prevalence, and prognosis among patients in the United States.

METHODS

Search strategy

We searched Ovid MEDLINE, Ovid MEDLINE In-Process, Ovid EMBASE, PubMed, and the Cochrane Library from inception through September 2023, using search terms described in the Supplemental Methods, http://links.lww.com/HC9/A835. A manual search of references from relevant articles was performed to identify publications missed by search terms. A manual search of the American Association for the Study of Liver Diseases (AASLD), the European Association for the Study of the Liver, Digestive Diseases Week, and the American College of Gastroenterology meeting abstracts from 2019 to 2021 was also completed. This study was conducted in accordance with the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) guidelines.¹⁹

Study selection and inclusion/exclusion criteria

The inclusion criteria were as follows: (1) cohort, cross-sectional, or case-control trials including original data characterizing racial/ethnic differences among adult patients (≥18 y) with ALD in the United States, (2) ALD (defined as any patient with AFL or AC or AH) diagnosed using biochemical, radiologic, or histologic criteria per AASLD guidelines, and (3) reported at least one of the following: ALD prevalence, severity, or prognosis. We excluded studies that (1) did not stratify ALD prevalence, severity, or prognosis by race/ethnicity, (2) had insufficient data to determine the necessary denominator for prevalence or severity, or (3) included patients with other causes of hepatic steatosis (eg, nonalcoholic liver disease, medication-induced, or HIV infection). Additional exclusion criteria included: (1) lack of original data (eg, editorials, review articles), (2) nonhuman studies; (3) incomplete data, and (4) non-English language. For studies with overlapping cohorts, articles with the most contemporary cohort or complete data were chosen. Studies analyzed Hispanic ethnicity as a mutually exclusive group without considering further division by race.

Data extraction and quality assessment

Studies were screened and reviewed in a collaborative, multistep process. After removing duplicates, 2 investigators (Thomas G. Cotter and Jeremy Louissaint) independently reviewed publications identified by the search strategy. Articles were screened based on title and abstract for relevance, followed by full-text review to assess for inclusion. Disagreements between authors were resolved by discussion with a third reviewer (Amit G. Singal). Using standardized forms, 2 authors (Thomas G. Cotter and Jeremy Louissaint) independently extracted data including patient demographics (including race/ethnicity), method of ALD diagnosis (AFL, AH, and AC), ALD prevalence, ALD severity, and ALD prognosis outcomes (including liver-related and all-cause mortality). Study quality was assessed using a modified checklist based upon the NIH Quality Assessment Tool for Observational Cohort and Cross-sectional Studies, which rates observational studies on a 14-point scale based on study sample appropriateness, comparability of study groups, and adequacy of assessing exposure and outcomes assessed by 2 investigators (Ahmad Anouti and Thomas G. Cotter).²⁰

Statistical analysis

For each study, we calculated a pooled prevalence and pooled incidence of ALD, AH, and AC within the general population and inpatient cohorts. For mortality assessment, we extracted proportions of ALD deaths among all causes of mortality. We used the DerSimonian and Laird method for random-effect models to calculate pooled risk ratio (RR) estimates for prevalence and proportion of ALD (including AH and AC) and pooled OR estimates for mortality. We used the χ² test of heterogeneity and the inconsistency index (I ²) to quantitatively determine the extent of heterogeneity between studies. If significant heterogeneity (I ²>50%) was found between studies, we used sensitivity analysis with graphical display of study heterogeneity diagnostics to exclude outlier studies one at a time to assess if this impacted pooled effect estimates. We performed subgroup analyses to explore potential causes of heterogeneity among results. Potential publication bias was evaluated graphically using Begg funnel plot and statistically using Egger regression test and fail-safe N.²¹,²²,²³ All data analyses were performed using R software 4.2.1.

RESULTS

Literature search

The search strategy yielded 15,980 potentially relevant citations. After removing 3904 duplicate citations, 10,353 unique citations remained. After initial review, 267 titles met the inclusion criteria. Of 192 publications that underwent full-text review, 54 met the inclusion criteria. Six studies contained overlapping data, so we selected the most inclusive studies representing patients from nonoverlapping cohorts; therefore, 25 eligible studies were retained. Agreement between reviewers for final study inclusion exceeded 95%. Supplemental Figure S1, http://links.lww.com/HC9/A835 represents a flow diagram depicting study selection.

Study characteristics

We included 28 studies, with a total of 76,867,544 individuals, characterizing racial/ethnic differences in ALD prevalence, severity, or prognosis. Not all studies had a detailed breakdown of the individuals based on race or ethnicity. The studies that had a racial and ethnic breakdown had a composition including 41,930,457 (54.5%) White, 135,580 (0.2%) Hispanic, 2,060,506 (2.7%) Black, 96,866 (0.13%) Asian, 29,584 (0.04%) AI/Alaskan Native (AN), and 335,758 (0.4%) “other” race and ethnicity individuals. All studies reported races and ethnicities as mutually exclusive categories. Of the included studies, 6 characterized differences in ALD prevalence (5 studies among the general population and 1 among inpatient populations), 9 addressed ALD severity, and 11 ALD mortality; 3 studies reported disparities in more than 1 study outcome.

ALD prevalence and incidence in population-based cohorts

Five studies (n=336,593) assessed the prevalence of ALD in population-based cohorts (Table 1).¹³,¹⁶,²⁴,²⁵,²⁶ Across studies, ALD was determined by reports in the medical record of excessive alcohol use and elevated aspartate transaminase or alanine aminotransferase in the absence of viral hepatitis (B or C) and no hepatotoxic medications (n=2) and international classification of diseases (ICD) coding (n=3).

TABLE 1.

Studies characterizing racial/ethnic disparities in ALD prevalence, severity, and mortality among US patients

References	Study design	Cohort	Study period	Method of ALD diagnosis	n	n (Whites/Hispanics/Blacks/Asians/native Americans/Others)	n ALD (overall prevalence)	n ALD (Whites/Hispanics/Blacks/Asians/Native Americans/Others)
ALD prevalence population based
Doycheva et al16	Cross-sectional	Retrospective NHANES	1988–2012: split into 3 study periods (1988–1994, 1999–2004, 2007–2012)	Excessive alcohol use and elevated AST or ALT in the absence of viral hepatitis (B or C) and no hepatotoxic medications.	14,547	Not available as participants were population weighted	1988–1994: 2.30% (1.50–3.10)/1999–2004: 4.40% (3.50–5.40)/2007–2012: 5.10% (4.20–5.90)	Whites: 1988–1994: 2.2% (1.3–3.1)/1999–2004: 5.10% (3.80–6.50)/2007–2012: 5.40% (4.20–6.70) Hispanics: 1988–1994: 4.10% (2.90–5.30)/1999–2004: 3.80% (2.50–5.10)/2007–2012: 7.50% (6.10–8.90) Blacks: 1988–1994: 2.40% (1.60–3.20)/1999–2004: 2.30% (1.10–3.40)/2007–2012: 2.50% (1.50–3.40)
Setiawan et al13	Cross-sectional	Retrospective multiethnic cohort	1999–2012	ICD coding for CLD or cirrhosis, ALD defined as any coding with alcohol complication or heavy drinking on questionnaire	106,458	28,255/21,492/15,782/33,312/NA/7,647	1195 (1.1%)	345/347/160/276/NA/67
Wong et al24	Cross-sectional	Retrospective NHANES	2001–20016: we only assessed 2015–2016 study period	Excessive alcohol use, elevated liver enzymes, excluding Hep B/C	215,251	49,292/47,355/35,085/56,826/NA/NA	180 (0.08%)	61 (4.1% (3.2–5.8))/87 (9.3% (8.7–10.0))/22 (3.4% (1.7–4.0))/NA/NA/10 (2.7% (1.0–3.4))
Gonzalez et al25	Cross-sectional	Retrospective Henry Ford Health System	2016–2020	ICD-10- CM: Alcoholic hepatitis (K70.10/K70.11), Alcoholic hepatic failure (K70.40/K70.41), Alcoholic cirrhosis (K70.30/K70.31), and Alcoholic liver disease, unspecified (K70.9)	NA	NA	337	274/NA/44/NA/14/5
Fischer et al26	Cross-sectional	Retrospective EMR Alaska	2003–2004	ICD coding, elevated liver enzymes, chart review	26,166	NA/NA/NA/NA/26,166/NA	792 (3.0%)	Other: Men: 37.1 per 1000 and Women: 25.1 per 1000
ALD prevalence high-risk cohorts
Tao et al27	Cross-sectional	Retrospective hospital discharge data from California Office of Statewide Health Planning and Development	1999	ICD coding including alcoholic fatty liver (571.0) and unspecified alcoholic liver damage (571.3).	609,745	NA	2057 (0.3%)	782/1008/203/58/NA/NA
Alcoholic hepatitis (AH)
Levy et al28	Cross-sectional	Retrospective UC David Medical Center EMR	2002–2010	ICD-9 codes: AFLD = +US and elevated liver enzymes and alcohol abuse	791	493/159/50/NA/NA/89	154 (19.5%)	101/28/11/NA/NA/14
Shirazi et al29	Cross-sectional	Retrospective NIS	2007–2014	ICD-9 codes	159,973	96,395/24,823/12,913/1,427/3,034/NA	29,509	18,533/3196/2729/281/395/NA
Yang et al30	Cross-sectional	Retrospective NIS	1998 and 2004	ICD-9 codes	NA	NA	6.6 per 100,000	Per 100,000: 7.3/5.9/5.1/4.0/5.5/NA
Noronha et al31	Cross-sectional	Retrospective single-center UNMH	2010–2012	ICD-9 codes	11,383	5744/1684/471/NA/1262/2222	191 (1.7%)	48/53/2/NA/NA/65/23
Garagliano et al32	Cross-sectional	Retrospective General Hospitals in the Baltimore Standard Metropolitan Statistical Area	1970–1973	Histological with clinical findings and supportive lab findings.	2,128,161	NA	NA	White males: 44.9 per 100,000 White females: 19.1 per 100,000 Non-White males: 50.1 per 100,000 Non-White females: 25.9 per 100,000
Ladhani et al33	Cross-sectional	Retrospective NIS	2012–2016	ICD-9: 571.1 and ICD-10: K70.10	NA	NA	21,898	67.3%/11.6%/9.8%/NA/NA
Wakil et al34	Cross-sectional	Retrospective NIS	2011–2017	ICD-9 and ICD-10 codes	NA	NA	2,071,936	1,359,386/289,825/195,163/21,065/44,101/54,789
Damjanovska et al35	Cross-sectional	Retrospective population-based commercial database (Explorys Inc, Cleveland, OH)	1999–2021	Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT)	74,033,580	40,266,840/NA/1,922,670/NA/NA/NA	2019–June 21, 2020: 23,350 per 100,000 June 21, 2020–June 20, 2021: 8320 per 100,000	Whites: • 2019–June 21, 2020: 5870 per 100,000 • June 21, 2020–June 20, 2021: 18,590 per 100,000 Blacks: • 2019–June 21, 2020: 1910 per 100,000 • June 21, 2020–June 20, 2021: 2380 per 100,000
Alcoholic cirrhosis (AC)
Levy et al28	Cross-sectional	Retrospective UC Davis Medical Center EMR	2002–2010	ICD-9 codes: AFLD = +US and elevated liver enzymes and alcohol abuse	791	493/159/50/NA/NA/89	345 (43.6%)	311/113/30/NA/NA/53
Shirazi et al29	Cross-sectional	Retrospective NIS	2007–2014	ICD-9 codes	159,973	96,395/24,823/12,913/1427/3034/NA	133,929 (83.7%)	80,092/22,004/10,380/1174/2698/NA
Yang et al30	Cross-sectional	Retrospective NIS	1998 and 2004	ICD-9 codes	NA	NA	13.6 per 100,000	Per 100,000: 13.5/11.1/15.5/15.1/11.1/NA
Kilbourne et al36	Cross-sectional	Retrospective Medicare Tennessee	2000	ICD-9 codes	470,032	418,799/NA/51,233/NA/NA/NA	1663 all alcohol-associated diagnosis (0.4%)	362/NA/45/NA/NA/NA
Sofair et al37	Cross-sectional	Prospective population-based surveillance in gastroenterology practices Multnomah County, Oregon and New Haven County, Connecticut; and primary care and gastroenterology practices from Kaiser Permanente Northern California in Alameda County	1999–2001	Definite ALD was either: (1) liver pathology consistent with ALD and history of regular alcohol use within the last 2 y; or (2) heavy alcohol consumption for Z10 years and elevated liver test	1040	NA	82 (7.9%)	65 (79.3%)/NA/9 (11%)/3 (3.6%)/NA/5 (6.1%)
ALD mortality
Allouch et al38	Cross-sectional	Retrospective United Autoworkers-GM Cohort + National Death Index	1941–2015	ICD-9/10 codes	41,097	6388/26,507/NA/NA/NA/8802	442	69/NA/248/NA/NA/125
Moon et al17	Cross-sectional	Retrospective CDC WONDER	1999–2017	ICD-10 (all alcohol ALD (K70))	22,231 deaths	15,902/3484/1660/NA/871/NA	NA	Per 100,000: M: 13.19 (12.93–13.45) W: 6.34 (6.15–6.52)/ M: 19.04 (18.3–19.78) F: 4.23 (3.91–4.55)/ M: 8.29 (7.78–8.8) F: 3.74 (3.43–4.05)/NA/ M: 34.11 (31.02–37.21) F: 24.48 (21.85–27.10)/NA
Woolf et al15	Cross-sectional	Retrospective CDC WONDER	1999–2016	ICD-10 (all alcohol)	NA	NA	NA	Per 100,000: NA/10.4/5.2/2.0/48.6/NA
Mendenhall et al39	Cohort study	Prospective VA Multicenter	1978–1983	Biopsy (64%) and clinical diagnosis	NA	NA	437	256/63/109/NA/9/NA
Kposowa et al40	Cross-sectional	Retrospective National Longitudinal Mortality Study (NLMS)—US Census Bureau	1990–2011	ICD-9 (K70.1-4.9) codes	1,376,200	1,057,477/NA/NA/NA/NA/318,573	NA	NA
Hadland et al41	Cross-sectional	Retrospective CDC WONDER	2002–2011	ICD-10 (all alcohol ALD (K70))	NA	NA	NA	Per 100,000 4/5.6/4.1/NA/7.4
Hurwitz et al18	Cross-sectional	Retrospective CDC WONDER	1979–1989	ICD ALD	303,875 deaths			Per 100,000: 4.37/NA/8.61/NA/NA/4.55
Stinson et al42	Cross-sectional	Retrospective CDC	1991–1997	ICD-9 all 571 codes (ALD)	174,923 deaths from all cirrhosis causes	128,848/18,268/21,696/NA/NA/NA
Ladhani et al33	Cross-sectional	Retrospective NIS	2012–2016	AH ICD-9: 571.1 and ICD-10: K70.10 AC: Prior Published Algorithms	NA	NA	21,898	White: reference Hispanic: OR: 1.02 (0.75–1.37) Black: OR: 0.52 (0.34–0.79)
Wakil et al34	Cross-sectional	Retrospective NIS	2011–2017	ICD-9 and ICD-10 codes	NA	NA	2,071,936	White: reference Hispanic: OR: 0.94 (0.9–0.98) Black: OR: 0.82 (0.77–0.86) Asian: OR: 0.92 (0.80–1.07) Native Americans: OR: 1.04 (0.94–1.15) Other: OR: 1.12 (1.02–1.22)
Ilyas et al43	Cross-sectional	Retrospective NIS	1999–2022	ICD-10 codes	NA	NA	436,651	378,297/NA/36,095/15,927/5342/NA

Abbreviations: AFLD, alcohol-associated fatty liver disease; ALD, alcoholic-associated liver disease; ALT, alanine aminotransferase; AST, aspartate transaminase; AH, alcoholic-associated hepatitis; AC, alcoholic-associated cirrhosis; APC, annual percent change; ICD, international classification of diseases; NA, not available.

Three studies examined the pooled prevalence of ALD by race and ethnicity.¹³,¹⁶,²⁴ ALD prevalence ranged from 0.3% to 5.4% (pooled prevalence: 2.1%; 95% CI: 0.71–6.0) in White individuals, from 1.6% to 9.3% (pooled prevalence: 4.5%; 95% CI: 2.1–9.4) in Hispanic individuals, and from 0.15% to 3.4% (pooled prevalence: 1.4%; 95% CI: 0.5–4.1) in Black individuals (Figure 1). Compared to White individuals, the pooled RR of ALD prevalence among Hispanic individuals was 1.41 (95% CI: 0.83–2.37), although there was significant heterogeneity (I ²=100%). The pooled RR of ALD among Hispanics was significantly higher than White individuals (RR: 1.64; 95% CI: 1.12–2.39, I ²=99.9%) after removing the study by Doycheva et al,¹⁶ an outlier on forest plot analysis. The pooled RR of ALD prevalence between Black and White individuals was 0.66 (95% CI: 0.45–0.98, I ²=100%). After removing the study by Doycheva et al,¹⁶ the difference among Black and White individuals with ALD prevalence remained statistically significant (pooled RR: 0.55; 95% CI: 0.35–0.87, I ²=95.5%) (Figure 2). The study by Fischer et al²⁶ reported ALD prevalence within the AI/AN population. Among 26,166 AI/AN patients from the Alaskan Native Medical Center, 792 (3.0%) had ALD based on ICD coding, labs, and chart review.

FIGURE 1.

Pooled prevalence of ALD (%) from the general population among (A) White, (B) Hispanic, and (C) Black individuals. Abbreviation: ALD, alcohol-associated liver disease.

FIGURE 2.

Pooled RR of ALD prevalence: (A) Hispanic compared to White patients with the outlier included (Doycheva 1999–2004). The funnel plot highlights the study outlier and the statistically significantly increased RR among Hispanic compared to White patients without the outlier (RR: 1.64; 95% CI: 1.12–2.39, I ²=99.9%). (B) Black compared to White patients with the outlier study. The funnel plot indicates the study outlier and compares Black patients to White patients (pooled RR: 0.55; 95% CI: 0.35–0.87, I ²=95.5%). Abbreviations: ALD, alcohol-associated liver disease; RR, risk ratio.

Three studies (n=5,597,979) used ICD-9 codes to assess the differences among Black and White individuals with AC prevalence within the general population.³⁰,³⁶ The prevalence of AC in White individuals ranged from 11.1 to 81.6 per 100,000 versus 9.9 to 76.1 per 100,000 among Black individuals (Supplemental Figure S1, http://links.lww.com/HC9/A835). The pooled RR between Black and White individuals was 0.93 (95% CI: 0.77–1.12, I ²=0%) (Supplemental Figure S2, http://links.lww.com/HC9/A835). Only the study by Yang et al³⁰ described the prevalence of AC among Hispanic individuals in the general population which was 16.9 per 100,000.

Finally, only 1 study reported AC and AH incidence rates per 100,000 of the general population in 1973 and found lower AC incidence rates among White individuals compared with non-White individuals (44.9 in White males and 19.1 in White females vs. 50.1 in non-White males and 25.9 in non-White females) while non-White individuals had the highest AH incidence rate (11.6 in White males and 5.7 in White females vs. 33.2 in non-White males and 6.8 in non-White females).³²

ALD prevalence and incidence among inpatient cohorts

One retrospective study (n=609,745) assessed ALD prevalence among patients discharged from the hospital using California Office of Statewide Health Planning and Development data. Using ICD coding, the prevalence of ALD was 33.6 (95% CI: 31.9–35.3) per 100,000 in men and 10.7 (95% CI: 9.7–11.6) per 100,000 in women. In White patients, ALD prevalence was 25.7 (95% CI: 23.5–28) per 100,000 in men and 11.9 (95% CI: 10.4–13.4) per 100,000 in women. Compared to Whites, ALD prevalence was higher in Hispanic men (61.1 per 100,000, 95% CI: 56.6–65.7) but similar among Hispanic women at 11.7 (95% CI: 9.9–13.5) per 100,000. Black patients had similar ALD prevalence as White patients, at 24.8 (95% CI: 20.3–29.3) and 13.9 (95% CI: 10.8–16.9) per 100,000 in men and women, respectively. Asian men and women had the lowest ALD prevalence at 8.0 (95% CI: 5.7–10.3) and 1.1 (95% CI: 0.4–1.9) per 100,000, respectively.²⁷

ALD severity

Two studies (n=160,764) assessed the proportions of AC plus AH among patients with ALD,²⁸,²⁹ and 3 studies (n=161,804) assessed the proportions of AC among patients with ALD.²⁸,²⁹,³⁷ Shirazi and colleagues defined ALD using ICD-9 codes (which included AFL, ICD-9 code 571.0), while Levy and colleagues used a combination of ICD-9 codes with clinical imaging and laboratory values to define AFL, AH, and AC, and Sofair and colleagues defined ALD as liver pathology consistent with ALD and history of regular alcohol use within the last 2 years or heavy alcohol consumption for ≥10 years and elevated liver tests.

The pooled proportion of AC was 67.1% (95% CI: 26.7–91.9, I ²=99%) in White patients with ALD, compared to 81.9% (95% CI: 2.5–99.9, I ²=98%) and 59.9% (95% CI: 10.3–95.1, I²=91%) in Hispanic and Black patients with ALD, respectively (Figure 3). Compared to White patients, the pooled RR of AC in ALD was 1.07 (95% CI: 1.04–1.10, I ²=0%) and 0.97 (95% CI: 0.96–0.98, I ²=0%) among Hispanic and Black patients, respectively (Figure 4).²⁸,²⁹

FIGURE 3.

The proportion (%) of AC among patients with ALD: (A) White, (B) Hispanic, and (C) Black patients. Abbreviations: AC, alcohol-associated cirrhosis; ALD, alcohol-associated liver disease.

FIGURE 4.

Pooled RR of the proportion of AC among patients with ALD: (A) Hispanic compared to White patients and (B) Black compared to White patients. Abbreviations: AC, alcohol-associated cirrhosis; ALD, alcohol-associated liver disease; RR, risk ratio.

The pooled proportions of AH were 19.2% (95% CI: 17.7–20.9, I ²=0%) in White patients with ALD, 12.9% (95% CI: 10.4–15.9, I ²=68%) in Hispanic patients with ALD, and 21.1% (95% CI: 16.9–26.1, I ²=0%) in Black patients with ALD (Supplemental Figure S3, http://links.lww.com/HC9/A835). Compared to White patients, the pooled RR of the proportion of AH in ALD was 0.7 (95% CI: 0.2–2.5, I ²=40%) and 1.1 (95% CI: 1.08–1.12, I ²=0%) among Hispanic and Black patients, respectively (Supplemental Figure S4, http://links.lww.com/HC9/A835).²⁸,²⁹

A study by Noronha et al³¹ (n=11,383) assessed patients with severe AH (identified by ICD-9 coding and Maddrey Discriminant Function score >32) among all hospital admissions at a single hospital center in New Mexico. The proportion of patients with severe AH was 0.8%, 3.1%, 0.4%, and 5.2% for White, Hispanic, Black, and AI patients, respectively. Damjanovska and collaegues (n=74,033,580) assessed patients with several alcohol etiologies (identified by systemized nomenclature of medicine-clinical terms) among patients of a population-based commercial database (Explorys Inc.). White patients reported 5870 per 100,000 cases of AH from 2019 to June 21, 2020, and 18,590 per 100,000 AH cases from June 21, 2020 to June 20, 2021, while Black patients reported 1910 per 100,000 cases of AH from 2019 to June 21, 2020, and 2380 per 100,000 AH cases from June 21, 2020 to June 20, 2021.³⁵

ALD mortality

Seven studies (n=542,126) assessed racial/ethnic differences in the prognosis of ALD.¹⁵,¹⁷,¹⁸,³⁸,⁴¹,⁴²,⁴³ The pooled mortality among White and Black patients was 18.0 (95% CI: 0.41–786.88) and 21.8 (95% CI: 0.5–865.3) per 100,000 respectively. In comparison, pooled mortality was higher among Hispanics 9.9 (95% CI: 4.5–21.7) per 100,000. Ilyas and collaegues reported ALD mortality among AI/AN (28.5 per 100,000) and Asian (2.0 per 100,000) patients. Pooled mortality ratios did not significantly differ between White and Black patients (1.2; 95% CI: 0.8–1.6; I ²=0%), White and Hispanic patients (1.54; 95% CI: 0.9–2.5; I ²=0%), and White and AI/AN patients (1.6; 95% CI: 0.9–2.9). Pooled mortality ratios did significantly differ between White and Asian patients (0.1; 95% CI: 0.03–0.5) (Table 2) (Figure 5).

TABLE 2.

Mortality analysis: pooled proportion of mortality and comparison of mortality risk among patients with ALD stratified by race

Race	Pooled proportion	95% CI	OR	95% CI
White	18.0	0.4–786.9	Reference	Reference
Hispanic	9.9	4.5–21.7	1.5	0.9–2.5
Black	21.8	0.5–865.3	1.2	0.8–1.6
AI/AN	28.5	NA	1.6	0.9–2.9
Asians	2.0	NA	0.1	0.03–0.5

Abbreviations: AI/AN, American Indians/Alaskan Natives; NA, not available.

FIGURE 5.

The OR of ALD mortality using White patients as the reference: (A) Hispanic and (B) Black patients. Abbreviation: ALD, alcohol-associated liver disease.

Four additional studies reporting mortality among patients with ALD could not be included in pooled analyses. Mendenhall and colleagues (n=473) assessed mortality among patients with ALD within the VA health system. Using Black patients with ALD as the reference population, the authors found a higher risk of mortality among White (HR: 1.7, 95% CI: 1.2–1.5) and Hispanic (HR: 1.9, 95% CI: 1.1–3.1) patients.³⁹ Another study by Kposowa and colleagues (n=1,376,200) assessed patient mortality among patients with ALD using the National Longitudinal Mortality Study US Census Bureau. Using White patients as the reference population, non-White ALD minorities had a significantly increased risk of ALD mortality (HR: 1.55, 95% CI: 1.53–1.85).⁴⁰ Ladhani et al³³ assessed 21,898 ALD patient mortality using the NIS database. Using White patients with ALD as the reference population, the authors found no difference in Hispanic patients (OR: 1.02, 95% CI: 0.75–1.37) and decreased risk of mortality among Black patients (OR: 0.52, 95% CI: 0.4–0.79). Wakil and colleagues assessed 2,071,936 ALD patient admissions using the NIS database. Using White patients with ALD as the reference population, the authors found decreased risk in Hispanic patients (OR: 0.94, 95% CI: 0.9–0.98) and Black patients (OR: 0.82, 95% CI: 0.77–0.86), while no difference was found in Asian (OR: 0.92, 95% CI: 0.8–1.07) and AI/AN (OR: 1.04, 95% CI: 0.94–1.15).³⁴

Quality assessment

Quality assessment of studies is provided in Table 3. Most studies had appropriate cohort selection, including the representativeness of the inpatient cohort. The most common limitation observed was the cross-sectional study design (24 of 25 studies), which precluded the exposure being measured before the outcome, sufficient time frame, repeated measurements of the exposure over time, and loss to follow-up reporting. Although outcomes were clearly defined and valid in most studies, ascertainment methods for those outcomes varied. Some studies poorly defined their research question (4 of 25 studies). Moreover, studies that assessed ALD severity did so through assessing either prevalence or incidence of AH and AC within heterogeneous population groups. Finally, several studies used large administrative databases, for example, NHANES, NIS, and CDC WONDER, which have inherent limitations, including missing or incomplete data on the diagnosis of ALD thus increasing the risk of ascertainment bias, and several studies contained overlapping data leading to exclusions of many prognostic studies.

TABLE 3.

Quality assessment of studies included in the meta-analysis

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	Overall score
Doycheva et al16	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	Yes	NA	NA	7
Setiawan et al13	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	Yes	Yes	Yes	NA	No	8
Wong et al24	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	Yes	NA	Yes	8
Gonzalez et al25	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	5
Fischer et al26	Yes	Yes	Yes	Yes	NA	No	No	No	No	No	Yes	No	NA	No	5
Tao et al27	No	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	5
Levy et al28	Yes	Yes	Yes	Yes	NA	No	No	Yes	Yes	No	Yes	No	NA	No	7
Shirazi et al29	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Yang et al30	No	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	5
Noronha et al31	No	No	Yes	No	NA	No	No	No	No	No	Yes	No	NA	Yes	3
Garagliano et al32	No	Yes	Yes	Yes	NA	No	No	No	No	No	Yes	No	NA	No	4
Ladhani et al33	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Wakil et al34	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Damjanovska et al35	Yes	Yes	Yes	Yes	NA	No	No	No	No	No	No	No	NA	No	4
Kilbourne et al36	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Sofair et al37	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Allouch et al38	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Moon et al117	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Woolf et al15	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Mendenhall et al39	No	Yes	Yes	Yes	NA	Yes	Yes	No	Yes	Yes	Yes	No	NA	Yes	9
Kposowa et al40	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Hadland et al41	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7
Hurwitz et al18	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Stinson et al42	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	No	6
Ilyas et al43	Yes	Yes	Yes	Yes	NA	No	No	No	Yes	No	Yes	No	NA	Yes	7

Note: Overall score: Total score out of 14 for each paper using the NIH Quality Assessment Tool for Observational Cohort and Cross-sectional Studies.

Abbreviation: NA, not available.

DISCUSSION

Understanding the contemporary epidemiology of ALD prevalence, severity, and prognosis among racial and ethnic groups is important to guide future interventions and reduce disparities in ALD burden. Our systematic review highlights that Hispanic individuals have the highest prevalence of ALD as well as the proportion of patients with ALD with AC. Conversely, Black individuals have significantly lower ALD prevalence and proportion with AC. Although there were no significant differences in ALD mortality between racial and ethnic groups, Hispanic and AI patients had a trend toward higher mortality rates compared to White patients.

The increased burden of ALD among Hispanics is noteworthy given similar rates of AUD among Hispanic and White individuals.⁴⁴ A possible explanation could include the presence of certain genetic factors (eg, PNPLA3) among Hispanic individuals which may render them more prone to alcohol-associated fatty liver disease in the setting of at-risk alcohol consumption.⁴⁵,⁴⁶,⁴⁷ In addition, the nature of alcohol consumption (eg, types of beverages and increased binge drinking) may also play a role. For example, Black girls (aged 11–18) reported less alcohol use (specifically beer, liquor, and wine) compared to White girls of the same age group; however, White individuals had higher use of wine when compared to Black, Latin, and Asian individuals.⁴⁸,⁴⁹ Social determinants of health (SDOH) (eg, household income) may also contribute to ALD prevalence,⁵⁰ with the development of ALD shown to correlate with socioeconomic factors in countries outside the United States such as China, Denmark, and India.⁵¹,⁵²,⁵³ Further studies are needed to investigate the potential role of SDOH in explaining racial and ethnic differences in ALD prevalence, including differential levels of education, income, and geographic location.

We identified only one study that examined ALD prevalence in an inpatient cohort, but this study was limited by ALD ascertainment methods. The method used in this paper did not properly encompass all patients with ALD and instead was only able to collect a convenient sample population.²⁷ This is a gap in the literature in addressing the prevalence of ALD in inpatient cohorts despite the high burden of AUD among inpatients—25% of inpatients met the criteria for AUD in one study.⁵⁴ Moreover, it is important to expand to other high-risk groups such as those with AUD at substance abuse treatment centers who are also at considerably increased risk of ALD. Inpatients should be assessed routinely for AUD and subsequent ALD.

With regard to ALD severity, there were no racial or ethnic differences in AC prevalence among the general population. However, Blacks had significantly lower prevalence of AC prevalence among ALD populations compared to other racial/ethnic groups.²⁸,²⁹ This finding, in addition to the fact that Black patients also had the lowest ALD prevalence merits further consideration given the fact that Black individuals have similar rates of AUD to other racial and ethnic groups.⁴⁴ Moreover, the Black cohort within the study does not adequately sample or reflect the race/ethnicity of the subgroups. The exact explanatory mechanisms for the apparent reduced burden and severity of ALD remain to be elucidated but may be related to reduced burden of liver disease modifiers (eg, obesity, diabetes)⁵⁵ in Black patients with ALD or possibly to increased misclassification of ALD due to concomitant hepatitis C infections becoming the primary diagnosis.⁵⁶ Of note, in the inpatient cohort analysis, Black people had an increased ALD prevalence and were on par with White people. This increased ALD burden among Black people in the inpatient environment may be in part explained by a later diagnosis of ALD due to various SDOH factors and possibly provider biases.⁵⁷

Among patients with ALD, the prevalence of AH was found to be significantly increased in Black patients compared to White patients, although we suspect this may have been related to a skewed population in the substantially larger study (Shirazi et al) which had a disproportionate number of Black patients with AH to what may be expected.²⁸,²⁹ An inpatient study demonstrated that, compared to NHW patients, NHB (incidence RRs: 0.79, CI: 0.77–0.81, p<0.001) and Hispanic patients (0.66, CI: 0.65–0.68, p<0.001) had lower incidence RRs of AH.⁵⁸ Moreover, Black patients have been shown to have a reduced mortality in AH compared to White patients (adjusted odds ratio: 0.50; 95% CI: 0.32–0.78).⁵⁹

In terms of morbidity and mortality, ALD is the primary cause of liver-related mortality.⁶⁰,⁶¹ While Hispanic, Black, and AI/AN patients had higher rates of ALD mortality numerically, compared to White patients, the differences were not significant.¹⁵,¹⁷,¹⁸,³⁸,⁴¹,⁴²,⁶² Of note, our results help to dispel the long-held belief of the “Hispanic paradox” (ie, Hispanic patients have lower ALD mortality rates despite a higher burden of disease) with Hispanic patients having similar mortality rates to White patients.¹¹ The low numbers of AI/AN patients with ALD increased the risk of type II error in the comparative analysis with White patients; thus, we cannot exclude the possibility that a significant difference may actually exist. Therefore, the pooled proportion of ALD mortality rate of 23.85 per 100,000 AI/AN individuals is noteworthy as it is substantially higher than that in all other racial and ethnic groups. In addition, AI/AN patients had an OR of 2.41 with the upper limit of the 95% CI being 13.73 compared to White patients. This reinforces the existing literature which has demonstrated a 6-fold increase among AI patients in ALD mortality compared to White patients.¹⁵,⁶³ The AI/AN populations remain understudied with regard to ALD epidemiology and the mechanisms behind the disproportionate mortality. Finally, Asian individuals appeared to have significantly reduced ALD mortality rates compared with White individuals. Future research is needed to understand the mechanisms behind this discrepancy.

The studies included in our systematic review and meta-analysis have numerous intrinsic drawbacks due to the difficulty of assessing racial and ethnic health disparities. First, information on race and ethnicity is self-reported and this may not reflect biological determinations of race and ethnicity. Second, multiethnic individuals add another layer of challenges, as these individuals are generally forced into a single category. Third, among the considered studies, the definition of Hispanic individuals was considered to be a race, despite the term Hispanic referring to an ethnicity. Therefore, there is a need in the field to specify further the race of Hispanic individuals. Fourth, a race-based analysis fails to account for the social determinants of health known to drive differences in health outcomes. Race and ethnicity are known to highly correlate with socioeconomic status, so failing to account for the intersection of the 2 constructs may overestimate disparities attributed to race and ethnicity. Fifth, there were little data on racial and ethnic groups such as AI/ANs and Asians. Sixth, most included studies were cross-sectional with limitations in study quality, which can also impact pooled estimates. Seventh, there was substantial heterogeneity for pooled estimates, which could not be resolved in sensitivity or subgroup analyses. Finally, there was a paucity of alcohol consumption information, so the role of alcohol (both prior and ongoing exposure) in ALD prevalence, severity, and mortality is less clear. We used AC and AH as surrogates for ALD severity, given the fact that there were no ALD papers with reliable risk stratification (ie, fibrosis staging) and racial and ethnic information. Moreover, the earlier stage of ALD (ie, AFL) remains under-recognized in the field which limited our analysis.

Lessons can also be learned from the hepatitis C literature, a similar patient population to ALD with high rates of chemical dependency and subjects prone to decreased follow-up rates. The quality and quantity of hepatitis epidemiological studies far exceed the existing ALD literature, including racial and ethnic disparities.⁶⁴,⁶⁵,⁶⁶,⁶⁷,⁶⁸ Future research should focus on large-scale systematic surveys of the prevalence of ALD stratified by race and ethnicity while using widely accepted methods of diagnosis. Design of large-scale multicenter studies is required to assess possible patients with ALD early while allowing for long-term follow-up of these patients. Stratifying levels of drinking and alcohol exposure will allow early diagnosis of ALD among the general population. We have highlighted several gaps in the ALD literature based on 4 categories (identifying populations, data sources, study design, and characterizing health behaviors) with possible solutions for each category (Table 4).

TABLE 4.

Overview of gaps in the literature with proposed futured directions

Subject/topic	Gaps in the literature	Steps forward
Identifying populations	• Lack of granularity and inclusion of diverse populations	• Advocate for increased studies in this area providing incentives (eg, grant support)
	• No characterization of multiracial groups	• Refinement of racial and ethnic grouping (including subgrouping broad racial groups, eg, Eastern and South Asian patients)
	• Earlier stages of ALD (eg, AFLD) remain undetected	• Improved classification with differentiation between cirrhosis and early-stage ALD
	• Lack of data on fibrosis stage in ALD leading to inadequate study of ALD severity	• More in-depth risk stratification of patient through noninvasive or invasive means
Data sources	• Limited databases study population (ie, CDC Wonder database)	• Encourage the use of other large databases and the development of multicenter studies
	• Limited data sources on social determinants of health	• Collecting social determinant variables (eg education, income, work, area of living) while performing prospective multicenter studies
Study design	• Paucity of epidemiological studies on patients with ALD (including the earlier stages, eg, AFLD)	• More granular alcohol use collection both at baseline and longitudinally
	• Cross-sectional studies - unable to establish causality or incidence	• Promote the design of prospective longitudinal studies
Characterizing health behaviors	• Insufficient quantification of alcohol use (previous and ongoing alcohol exposures) resulting in insufficient knowledge of the role of alcohol in ALD prognosis	• Performing longitudinal studies with long-term collection of patient data and previous/ongoing alcohol use

Abbreviations: AFLD, alcohol-associated fatty liver disease; ALD, alcohol-associated liver disease.

This systematic review and meta-analysis is the first comprehensive data synthesis of the epidemiology of ALD stratified by race and ethnicity. Some of our key findings include increased burden of ALD among Hispanics, reduced ALD severity among Black patients, and increased ALD mortality among AI/AN patients. Perhaps the most important finding was the confirmation that the ALD field lacks high-quality epidemiological studies in ethnically diverse patient populations. Improved epidemiological studies in this area can help guide hepatology practice and health policy to reduce ALD morbidity and mortality.

Supplementary Material

hc9-8-e0409-s001.docx ^{(346.9KB, docx)}

AUTHOR CONTRIBUTIONS

Ahmad Anouti and Thomas G. Cotter—study concept and design; acquisition, analysis, and interpretation of data; and drafting and critical revision of the manuscript for important intellectual content. Karim Seif El Dahan—study analysis and interpretation. Amit G. Singal, Lisa B. VanWagner, and Julia Kozlitina—study design and critical revision of the manuscript for important intellectual content. Nicole E. Rich, Jeremy Louissaint, Mack C. Mitchell, William M. Lee, Sarah R. Lieber, Juan Pablo Arab, Luis Antonio Díaz, Bill Y. Zhang, Mausam J. Patel, Chanattha Thimphittaya, Dyanna L. Gregory, and Andrea C. King—critical revision of the manuscript for important intellectual content. All authors approved the final version to be published.

FUNDING INFORMATION

Thomas G. Cotter is supported by the American Association for the Study of Liver Diseases (AASLD) Clinical, Translational and Outcomes Research Award (CTORA) and National Institute for Alcohol Abuse and Alcoholism (NIAAA) K23AA031310 grant. William M. Lee is supported by U01 DK58369. Andrea C. King is supported by R01-AA013746, R01-AA025366, R21-AA029746, and R01-DA044210. Amit G. Singal’s research is supported by the National Institute of Health R01 MD012565 and CA256977. Lisa B. VanWagner is supported by the NIH grant R56 HL155093.

CONFLICTS OF INTEREST

Nicole E. Rich consults and advises AstraZeneca, Elevar, and Exelexis. Mack C. Mitchell advises and received grants from GlaxoSmithKline. He is employed and owns stock in Amygdala. He advises HepaTx and Prodigy Biotech. He received grants from Durect. William M. Lee is supported by research support from Intercept, Aurora, Gilead, Novo Nordisk, Alexion, Eiger, Camurus, and Lipocine, and consults for Forma, SeaGen, GSK, Karuna, and Cortexyme. Amit G. Singal has served as a consultant or on advisory boards for FujiFilm Medical Sciences, Exact Sciences, Glycotest, GRAIL, Freenome, Universal Dx, Roche/Genentech, AstraZeneca, Eisai, Bayer, Exelixis, and TARGET RWE. Lisa B. VanWagner consults for Numares and Novo Nordisk and receives grant support from W.L. Gore & Associates and AstraZeneca, and serves as an expert witness. The remaining authors have no conflicts to report.