Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 9.
Published in final edited form as: Liver Transpl. 2025 Jun 9;31(12):1472–1487. doi: 10.1097/LVT.0000000000000653

Access to Early Liver Transplantation is Adversely Impacted by Social Determinants of Health: A Retrospective Cohort Study

James T Flanary 1,2, Po-Hung Chen 3,4, Sarah Shan 1,5, Jonathan Mitchell 1,8, Ahmet Gurakar 3, Alexandra T Strauss 3, Marie Diener-West 6, Michael R Desjardins 7, Sharon R Weeks 1,2, Kayleigh Herrick-Reynolds 1,2, Mayan Teles 1, Sumeyye Yilmaz 1, Daniel Warren 1, Elizabeth King 1, Andrew Cameron 1
PMCID: PMC12258191  NIHMSID: NIHMS2087699  PMID: 40476755

Abstract

Early liver transplant (ELT) is the practice of liver transplant for those with severe alcohol-associated hepatitis or acute on chronic alcohol-associated liver disease, without requiring any minimum pre-transplant abstinence period. It is an increasingly adopted practice for alcohol-associated liver disease (ALD) capable of providing excellent outcomes, but there are concerns regarding equity in access to ELT. Our objectives were 1) to quantify the association between social determinants of health (SDoH) and progression from referral to listing, and 2) to identify geographic disparities in referrals for ELT. We included 501 ELT candidates and a comparison population of 165 SLT candidates referred from 2017 until 2023. Multivariable regression quantified the association between individual- and neighborhood-level SDoH and progression from referral to listing, adjusting for sociodemographic characteristics and MELD. Negative binomial regression quantified the association between neighborhood-level SDoH and the number of ELT referrals within 150Nmi of our institution, adjusting for relevant geographic confounders. ELT candidates with Medicaid insurance were less likely to progress from referral to evaluation than candidates with private insurance (RR 0.69, 95% CI 0.56–0.85). ELT candidates from census tracts with a high social vulnerability index (SVI) were less likely to progress from evaluation to listing than candidates from low SVI census tracts in Modified Poisson Regression (RR 0.64, 95% CI 0.46–0.88), but not in Cox regression (HR 0.67, 95% CI 0.41–1.09). These findings were not significant among SLT candidates. High SVI census tracts were less likely to have individuals referred for ALD compared to tracts with low SVI (IRR 0.53, 95% CI 0.42–0.66). Both individual- and neighborhood-level SDoH are important factors that ultimately influence who is eligible for transplant.

Graphical Abstract

graphic file with name nihms-2087699-f0004.jpg

Introduction

Historically, transplants for alcohol-associated liver disease (ALD) have been performed only after an arbitrary 6 months of pre-transplant abstinence from alcohol.1 There is little evidence to suggest that requiring 6 months of pre-transplant abstinence reduces the risk of post-transplant alcohol relapse or improves relapse-free survival.2,3 For a subset of individuals that present with severe alcohol-associated hepatitis (SAH), the risk of mortality within 6 months is very high and precludes their ability to wait for transplants.4,5 Most centers now perform liver transplants for SAH with no minimum time requirement for pre-transplant abstinence in individuals who will otherwise die without a transplant and who meet other center-specific criteria.6 Several large reports describe experience with early liver transplants (ELT) for SAH, noting excellent post-transplant survival and similar rates of alcohol relapse as compared to recipients with ALD transplanted after 6-months of abstinence (standard liver transplantation, or SLT).7–9 The focus of initial ELT efforts has been on selecting appropriate recipients to ensure excellent post-transplant survival and minimize alcohol use relapse. The result has been a relatively homogenous group of recipients with regard to race, sex, and insurance status.10–12 As the practice of ELT is implemented more broadly, it is important that access to ELT be equitable and available to the entire population at risk for SAH.

Social determinants of health (SDoH) are important predictors of healthcare outcomes within the general population and within transplant populations.13,14 Disadvantages in SDoH have been associated with decreased transplant access, increased waitlist mortality, and inferior post-transplant outcomes.13,15–18 Individuals with SAH who lack access to substance use treatment and liver disease specialists may be at a distinct disadvantage in accessing ELT.19 In general, candidates referred and ultimately listed for liver transplant for all indications are less commonly female, Black, of lower socioeconomic status, or higher social vulnerability(24,25).15,16,20–22 It is likely that these disparities and others exist among candidates for ELT and may be exacerbated by the additional sociodemographic disparities in alcohol use disorder (AUD).12 Evaluating differences in sociodemographic characteristics through the referral process for ELT may provide insight into the impact of existing disparities and inform future outreach efforts to improve access to ELT and support for diverse populations.11,23 This study focuses specifically on the setting of ELT at the center with the largest cohort undergoing transplant for this indication.

To better understand disparities in ELT, the objective of this study was to quantify the association between SDoH and progression from referral to listing for candidates referred for ELT, relative to candidates referred for SLT. A secondary objective was to define whether there are geographic disparities in patient referrals for this indication.

Patients and Methods

Study population

The study population included 666 adult (≥18-year-old) candidates referred to our institution for ELT or SLT from January 1, 2017 through January 17, 2023, followed from the point of referral until their endpoint (i.e., either they received liver transplant or were considered an ineligible candidate). Internal (from providers within our healthcare system) and external referrals were included. Each candidate had ALD as a primary reason for their liver failure. ELT versus SLT was determined based on whether the candidate had 6 months of abstinence from alcohol at the time of referral. Candidates were excluded if they had a concomitant cause of liver failure (n=81). There were 148 individuals excluded because of at least one missing exposure variable. The majority of excluded individuals were referred, but not transferred for evaluation (n=107, 73%). ELT candidates that were excluded were younger, more frequently of non-white race, had Medicaid insurance, higher MELD, and higher prevalence of ongoing substance abuse, whereas SLT candidates that were excluded were more frequently of non-white race with no significant differences seen in other demographic categories (Supplement 1). The decision was made to omit imputation of missing data, because analysis with imputed data may lead to greater bias than a complete case analysis when data is missing not at random.26 The study abided by the Declarations of Helsinki and Istanbul, adhered to the STROBE cohort study reporting guidelines, and was approved by the institutional review board at Johns Hopkins School of Medicine (IRB00154881 and IRB00236315).27

Exposure Variables

Individual-level demographic, clinical, and SDoH variables were ascertained from electronic medical records (EMR) and a prospectively maintained database of candidate referrals for ALD. Individual-level variables included age; sex; race; ethnicity; type of insurance; marital status; distance to the transplant center; MELD score at time of referral, listing, and evaluation; presence of psychiatric comorbidities requiring additional management prior to transplant based on first evaluation by a social worker; and adequacy of social support at first evaluation by a social worker. Race and ethnicity data in our EMR were either patient-reported or perceived by the clinical staff.

Candidate addresses were geocoded using ArcGIS Pro to determine the candidate’s census tract and were linked to neighborhood-level variables.28 Candidates with a PO Box were geocoded based on city, state, and ZIP (n = 18, 3%). Neighborhood SDoH included the social vulnerability index (SVI) and its components (see below), census tract population aged 18 and over, mortality rates from ALD, the prevalence of AUD, and the prevalence of excessive drinking. ALD mortality rates were obtained at the county level from the Centers for Disease Control and Prevention, by ICD-10 code K.70, aggregated over the years 1999–2020, similar to previous studies.29–33 The percentage of the population over 18 with AUD was obtained from the National Survey on Drug Use and Health public use data at the substate region level for the years 2016–2018.34 The percentage of the population reporting excessive drinking was obtained at the county level from the County Health Rankings.24 To estimate county and substate rates at the census tract level, areal interpolation was used.35 This technique uses counts, population denominators, and geographic boundaries to estimate rates throughout the study area, allowing for reaggregation at other geographic levels.

We used the 2020 SVI, an index of 15 factors from the 2015–2020 American Community Survey. The SVI was developed to assist public health officials in resource allocation during an emergency response and is validated as a predictor of health outcomes. It includes four components: socioeconomic status, household composition and disability, minority status and language, and housing and transportation. These four components are used to calculate an overall index.36 The SVI has been used in transplant research to measure community-level SDoH in the assessment of equity.37,38 SVI was grouped by tertile of our sample into low, moderate, or high level of social vulnerability. In individual-level analysis, these thresholds were 0 – 20, 21 – 49, and 50 – 100. In geographic analysis, these thresholds were 0 – 26, 27 – 62, and 63 – 100.

Outcomes Variables

Progression from Referral to Listing

Referral was defined as an entry in our referral database or a healthcare provider initiating a transplant episode within our EMR, regardless of whether the candidate received medical care at our institution. Evaluation was defined as the point in which a candidate evaluation was initiated, regardless of whether it was fully completed. Listing was defined as the point in which a candidate was placed onto the national waiting list for transplant (Supplement 2).

Referrals by Census Tract

We evaluated all census tracts within a 150Nmi radius of our transplant center. Geographic boundaries of census tracts and counties were obtained from the 2020 administrative boundaries.39 This area corresponds to the majority of transplant referrals at our institution and also represents the nearest zone of referrals according to the United Network for Organ Sharing policy implemented in January, 2020.40 Counts of referred candidates for ELT and SLT within each census tract were calculated.

Statistical Analyses

Progression from Referral to Listing

Comparisons of candidate sociodemographic, clinical, and SDoH variables across stages of evaluation were performed with Kruskal-Wallis for continuous variables. Chi-square and Fisher’s exact tests were used for categorical variables, depending on whether all subgroups were greater than 5 (Table 1).

Table 1.

Demographic characteristics of ELT and SLT candidates stratified by stage of evaluation.

Early Liver Transplant Candidates Standard Liver Transplant Candidates
Total Referred, not evaluated Evaluated, not listed Listed, not transplanted Transplanted p-value Total Referred, not evaluated Evaluated, not listed Listed, not transplanted Transplanted p-value
a) Individual Level
Total (n) 501 195 155 33 118 165 14 45 43 63
Age, median [IQR] 49 [39, 58] 47 [38, 57] 51 [41, 60] 56 [47, 60] 49 [40, 57] 0.01 57 [49, 62] 54 [45, 61] 55 [48, 61] 54 [47, 62] 59 [51, 64] 0.15
Sex, n (% male) 294 (59) 102 (52) 94 (61) 22 (67) 76 (64) 0.11 118 (72) 9 (64) 34 (76) 28 (65) 47 (75) 0.60
Race, n (%) 0.11 0.51
 White 406 (81) 150 (77) 123 (79) 31 (94) 102 (86) 138 (84) 12 (86) 40 (89) 36 (84) 50 (79)
 Black 58 (12) 30 (15) 18 (12) 2 (6) 8 (7) 9 (5) 0 (0) 2 (4) 4 (9) 3 (5)
 Other 37 (7) 15 (8) 14 (9) 0 (0) 8 (7) 18 (11) 2 (14) 3 (7) 3 (7) 10 (15.9)
Type of Insurance, n (%) <0.001 0.69
 Private 274 (55) 90 (46) 84 (54) 20 (61) 80 (68) 97 (58) 6 (43) 29 (64) 24 (56) 38 (60)
 Medicare 77 (15) 22 (11) 28 (18) 7 (21) 20 (17) 34 (21) 5 (36) 6 (13) 10 (23) 13 (21)
 Medicaid 150 (30) 83 (43) 43 (28) 6 (18) 18 (15) 34 (21) 3 (21) 10 (22) 9 (21) 12 (19)
MELD, referral, median, [IQR] 29 [22, 36] 33 [25, 38] 25 [17, 30] 28 [22, 31] 33 [26, 38] <0.001 17 [12, 22] 22 [14, 30] 17 [12, 22] 14 [10, 19] 18 [14, 23] 0.03
MELD, listing, median, [IQR] 19 [12, 27] 30 [27, 35] <0.001 12 [8, 17] 19 [14, 25] <0.001
MELD, transplant, median, [IQR] 23 [19, 28] 19 [16, 24]
Psychiatric Comorbidities Requiring Additional Management, n (%) 31 (20) 2 (6) 7 (6) 0.001 9 (20) 5 (12) 7 (11) 0.37
Inadequate Social Support, n (%) 26 (17) 0 (0) 7 (6) 0.002 3 (7) 0 (0) 3 (5) 0.25
b) Geographic Level
In-state Residence 344 (69) 128 (64) 117 (75) 23 (70) 76 (64) 0.16 102 (62) 8 (57) 28 (62) 31 (72) 35 (56) 0.38
Social Vulnerability Index, n (%) 0.01 0.20
 Low Vulnerability 176 (35) 59 (30) 49 (32) 17 (52) 51 (43) 49 (30) 5 (36) 9 (20) 16 (37) 19 (30)
 Moderate Vulnerability 174 (35) 67 (34) 52 (34) 11 (33) 44 (37) 46 (28) 4 (29) 14 (31) 6 (14) 22 (35)
 High Vulnerability 151 (30) 69 (35) 54 (35) 5 (15) 23 (20) 70 (42) 5 (36) 22 (49) 21 (49) 22 (35)
Open in a new tab

Modified Poisson regression with robust variance estimation was used to estimate the relative risk of progression from referral to evaluation and progression from evaluation to listing.41 Cox proportional hazard models were used to estimate the hazard ratio for time to progression from referral to evaluation and progression from evaluation to listing. We estimated these effects separately for ELT and SLT candidates and also explored the interaction between ELT/SLT status and census tract SVI on progression from evaluation to listing. For the Cox proportional hazard model, time from referral to evaluation was administratively censored at 40 days after the date of referral, at which point a decision regarding evaluation was made for 93% of candidates. Time from evaluation to listing was administratively censored at 100 days after the date of evaluation, at which point a decision regarding listing was made for 93% of candidates. The proportional hazards assumption was not met beyond these intervals. All analyses were performed using R version 4.2.3.42

To investigate the effect of Medicaid on residents living both in state and out of state, we explored the interaction between a candidate’s state of residence and their insurance status and its effect on progression from referral to evaluation.

Referrals by Census Tract

Negative binomial regression quantified the association between census-tract SVI and number of ALD referrals at each census tract. See ‘Results‘ below for final model. Negative binomial was selected over Poisson regression due to overdispersion. It is possible that the distribution of the outcome is concentrated in certain areas, known as spatial correlation, which can potentially violate the assumption of independent observations.. The Moran’s I statistic was used to evaluate spatial correlation on the regression residuals. This value ranges between −1 (assumes complete spatial repulsion) to 1 (assumes complete spatial clustering), where 0 represents complete spatial independence. If the Moran’s I value is significantly different than 0, this suggests the assumption of independent sampling may be violated.43

A sensitivity analysis was performed to evaluate adjustment for alcohol-associated variables using both estimations from areal interpolation and from attributing reported county values to all census tracts within each county. Models using estimates via areal interpolation were selected based on the Akaike Information Criterion.44

Results

Individual-Level Analysis Comparing Stages of Evaluation

There were 501 candidates referred for ELT and 165 candidates referred for SLT that met the inclusion criteria. Among ELT candidates, the overall median age was 49 (IQR: 39–58), 59% were male, 81% were White, 55% had private insurance, and the median MELD was 29 (IQR: 22–36). Among SLT candidates, the overall median age was 57 (IQR: 49–62), 72% were male, 84% were white, 58% had private insurance, and the median MELD was 17 (IQR: 12–22) (Table 1). When compared to candidates who did not progress past the referral stage, ELT candidates who were either listed or transplanted had a higher proportion of being privately insured, a lower proportion of residing in census tracts with high social vulnerability, and a lower proportion of relevant psychiatric comorbidities. These differences in insurance type, SVI, and psychiatric comorbidities were not seen across stages of evaluation in the SLT candidates. There were no significant differences by sex or race in either the ELT or SLT cohorts.

Two of the final adjusted models included the outcome of progression from referral to evaluation, and included the exposure variables of age, sex, race, type of insurance, SVI, distance to transplant center, and MELD at time of referral. The remaining two final adjusted models included the outcome of evaluation to listing, and included the exposure variables of age, sex, race, type of insurance, SVI, presence of comorbid psychiatric disorder, adequacy of social support, distance to transplant center, and MELD at time of referral. Ethnicity and marital status were excluded from the final models due to a higher proportion missing and because these variables were found to be noncontributory on multivariable analyses.

Progression from Referral to Evaluation – Estimation of Relative Risk via Modified Poisson

In unadjusted analysis of the ELT cohort, candidates were less likely to progress from referral to evaluation if they were female versus male (RR 0.85, 95% CI 0.73–0.98) or insured with Medicaid versus privately insured (RR 0.66, 95% CI 0.55–0.81). After adjusting for other candidate characteristics, this difference by sex was no longer statistically significant (RR 0.90, 95% CI 0.78–1.04), but the difference by type of insurance remained statistically significant (RR 0.69, 95% CI 0.56–0.85). There was no evidence of a significant association between age, sex, race, distance to transplant center, or SVI and progression from referral to evaluation.

In unadjusted analysis of the SLT cohort, black candidates were more likely to progress from referral to evaluation than white candidates (RR 1.09, 95% CI 1.02–1.15). This difference did not persist after adjusting for other candidate characteristics. There was no evidence of a significant association between age, sex, insurance type, distance to transplant center, or SVI (Table 2).

Table 2.

Relative risk for progression from referral to evaluation and progression from evaluation to listing.

Early Liver Transplant Candidates Standard Liver Transplant Candidates
Progression from Referral to Evaluation Progression from Evaluation to Listing Progression from Referral to Evaluation Progression from Evaluation to Listing
Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis
Relative Riska (95% CI) Relative Riskb (95% CI) Relative Riska (95% CI) Relative Riskb (95% CI) Relative Riska (95% CI) Relative Riskb (95% CI) Relative Riska (95% CI) Relative Riskb (95% CI)
Age (10 year intervals) 1.04 0.97–1.11 1.30 0.92–1.06 1.05 0.94–1.16 0.99 0.89–1.10 1.00 0.95–1.06 1.02 0.96–1.08 1.06 0.95–1.19 1.00 0.88–1.13
Sex (ref = male) 0.85 0.73–0.98 0.90 0.78–1.04 0.95 0.76–1.20 1.08 0.86–1.36 0.97 0.87–1.09 0.99 0.89–1.10 1.07 0.86–1.34 1.13 0.91–1.40
Race (ref = white)
 Black 0.77 0.59–1.01 0.89 0.66–1.19 0.66 0.41–1.07 0.82 0.50–1.36 1.09 1.02–1.15 1.06 0.98–1.16 1.14 0.79–1.65 1.26 0.85–1.85
 Other 0.99 0.75–1.29 1.08 0.81–1.45 0.64 0.38–1.07 0.88 0.53–1.46 0.96 0.81–1.14 0.97 0.82–1.13 1.19 0.91–1.56 1.24 0.92–1.66
Type of Insurance (ref = private)
 Medicare 1.00 0.85–1.18 1.05 0.89–1.26 1.01 0.76–1.35 1.19 0.89–1.58 0.90 0.78–1.05 0.90 0.77–1.04 1.16 0.92–1.47 1.20 0.92–1.56
 Medicaid 0.66 0.55–0.81 0.69 0.56–0.85 0.69 0.50–0.96 0.83 0.60–1.15 0.98 0.87–1.10 0.97 0.85–1.12 0.99 0.75–1.32 1.18 0.87–1.6
SVI (ref = low)
 Moderate Vulnerability 0.93 0.79–1.08 0.95 0.82–1.11 0.88 0.70–1.11 0.88 0.70–1.09 1.02 0.91–1.16 1.03 0.90–1.17 0.84 0.64–1.09 0.80 0.62–1.04
 High Vulnerability 0.82 0.69–0.98 0.89 0.74–1.07 0.59 0.43–0.81 0.64 0.46–0.88 1.02 0.90–1.15 1.03 0.90–1.18 0.83 0.66–1.05 0.77 0.59–1.00
Distance to center(per 50 Nmi) 0.99 0.96–1.02 0.99 0.96–1.02 1.01 0.99–1.03 1.01 0.99–1.03 1.00 0.98–1.01 1.00 0.98–1.01 1.01 1.01–1.02 1.01 1.00–1.02
Psychiatric Comorbidities Requiring Additional Management 0.51 0.29–0.90 0.58 0.33–1.01 0.79 0.54–1.16 0.73 0.49–1.08
Inadequate Social Support 0.45 0.23–0.89 0.54 0.26–1.11 0.70 0.31–1.57 0.61 0.29–1.31
MELD (5 point interval) 0.91 0.88–0.95 0.92 0.89–0.95 1.25 1.16–1.33 1.22 1.14–1.31 0.96 0.92–1.01 0.96 0.92–1.01 1.00 0.93–1.08 0.99 0.92–1.07
Open in a new tab

When including an interaction between in-state residency status and insurance status, in addition to adjustment for other candidate characteristics, ALD candidates (i.e., both ELT and SLT candidates) residing in Maryland remained less likely to progress from referral to evaluation if they were insured with Medicaid versus privately insured (RR 0.0.84, 95% CI 0.72–0.98). Out-of-state candidates with Medicaid insurance were significantly less likely than in-state candidates with Medicaid insurance to progress from referral to evaluation, relative to those with private insurance (RR 0.30, 95% CI 0.12–0.75).

Progression from Evaluation to Listing – Estimation of Relative Risk via Modified Poisson

In unadjusted analysis, ELT candidates were less likely to progress from evaluation to listing if they were insured with Medicaid versus privately insured (RR 0.69, 95% CI 0.50–0.96), if they resided in census tracts with high SVI versus those with low SVI (RR 0.59, 95% CI 0.43–0.81), if they had relevant psychiatric comorbidities (RR 0.51, 95% CI 0.29–0.90), or if they had inadequate social support (RR 0.45, 95% CI 0.23–0.89). After adjusting for other candidate characteristics, the differences by insurance status, psychiatric comorbidities, and social support were no longer statistically significant, but the difference by SVI remained statistically significant (RR 0.64, 95% CI 0.46–0.88). There was no evidence of a significant association between age, sex, race, or distance to transplant center, and progression from evaluation to listing.

In the SLT cohort, further distance to the transplant center was associated with greater progression from evaluation to listing in both unadjusted and adjusted analysis (RR 1.01, 95% CI 1.01–1.02, and RR 1.01, 95% CI 1.00–1.02, respectively). There was no significant association between age, sex, race, insurance type, SVI, psychiatric comorbidities, or social support and progression from evaluation to listing (Table 2).

When including an interaction between ELT/SLT status and SVI, in addition to adjustment for other candidate characteristics, high SVI ELT candidates were significantly less likely than high SVI SLT candidates to progress from evaluation to listing, relative to low SVI candidates (RR 0.66, 95% CI 0.44–0.97).

Progression from Referral to Evaluation – Estimation of Hazard Ratio via Cox Regression

The median time from referral to evaluation for ELT candidates was 27 days (95% CI 21–35), and for SLT candidates, was 17 days (95% CI 14–19).

In unadjusted analysis of the ELT cohort, candidates were less likely to progress from evaluation to listing if they were female versus male (HR 0.76, 95% CI 0.60–0.97), black versus white (HR 0.65. 95% CI 0.43–0.99), insured with Medicaid versus privately insured (HR 0.57, 95% CI 0.43–0.77), or if they resided in census tracts with high SVI versus those with low SVI (HR 0.68, 95% CI 0.51–0.92). After adjusting for other candidate characteristics, the differences by sex, race, and SVI were no longer statistically significant, but the difference by type of insurance remained statistically significant (HR 0.62, 95% CI 0.45–0.85). There was no evidence of a significant association between age or distance to transplant center and progression from referral to evaluation for ELT candidates.

In unadjusted analysis of the SLT cohort, there was no significant association between age, sex, race, type of insurance, SVI, or distance to transplant center and progression from referral to evaluation. However, after adjusting for other candidate characteristics, those insured with Medicare versus private insurance were less likely to progress from referral to evaluation (HR 0.58, 95% CI 0.35–0.96) (Table 3).

Table 3.

Hazard ratio for progression from referral to evaluation and progression from evaluation to listing.

Early Liver Transplant Candidates Standard Liver Transplant Candidates
Progression from Referral to Evaluation Progression from Evaluation to Listing Progression from Referral to Evaluation Progression from Evaluation to Listing
Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis Unadjusted Analysis Adjusted Analysis
Hazard Ratioa (95% CI) Hazard Ratiob (95% CI) Hazard Ratioa (95% CI) Hazard Ratiob (95% CI) Hazard Ratioa (95% CI) Hazard Ratiob (95% CI) Hazard Ratioa (95% CI) Hazard Ratiob (95% CI)
Age (10 year intervals) 1.04 0.93–1.16 0.93 0.82–1.06 1.04 0.90–1.21 0.99 0.84–1.17 1.08 0.90–1.30 1.13 0.91–1.41 1.40 1.06–1.83 1.23 0.88–1.73
Sex (ref = male) 0.76 0.60–0.97 0.8 0.63–1.03 0.95 0.67–1.35 1.12 0.78–1.60 0.99 0.67–1.47 1.15 0.76–1.73 0.81 0.48–1.39 1.08 0.60–1.92
Race (ref = white)
 Black 0.65 0.43–0.99 0.77 0.49–1.20 0.70 0.36–1.35 0.87 0.43–1.76 1.33 0.62–2.82 1.34 0.61–2.94 1.24 0.480–3.13 1.79 0.66–4.87
 Other 0.88 0.55–1.42 0.93 0.57–1.52 0.57 0.28–1.17 0.83 0.38–1.81 1.21 0.70–2.09 1.29 0.73–2.29 1.07 0.51–2.26 1.02 0.45–2.32
Type of Insurance (ref = private)
 Medicare 1.09 0.79–1.50 1.27 0.90–1.82 0.96 0.62–1.51 1.11 0.68–1.83 0.64 0.40–1.03 0.58 0.35–0.96 1.09 0.61–1.95 1.01 0.54–1.90
 Medicaid 0.57 0.43–0.77 0.62 0.45–0.85 0.68 0.43–1.07 0.83 0.51–1.36 0.74 0.47–1.17 0.74 0.45–1.20 0.61 0.32–1.19 1.15 0.55–2.41
SVI (ref = low)
 Moderate Vulnerability 0.91 0.69–1.19 0.98 0.74–1.29 0.92 0.63–1.34 0.90 0.61–1.32 0.96 0.61–1.52 1.00 0.62–1.62 0.54 0.29–0.99 0.58 0.30–1.12
 High Vulnerability 0.68 0.51–0.92 0.76 0.55–1.04 0.61 0.39–0.96 0.67 0.41–1.09 0.87 0.57–1.32 0.95 0.61–1.50 0.60 0.35–1.04 0.60 0.32–1.10
Distance to center(per 50 Nmi) 0.98 0.95–1.02 0.97 0.93–1.01 1.02 0.98–1.06 1.01 0.97–1.05 0.98 0.95–1.02 0.98 0.94–1.02 1.15 1.06–1.25 1.13 1.05–1.23
Psychiatric Comorbitidies Requiring Additional Management 0.37 0.17–0.79 0.42 0.19–0.94 0.35 0.14–0.88 0.36 0.14–0.94
Inadequate Social Support 0.37 0.17–0.80 0.44 0.20–0.95 0.26 0.04–1.87 0.25 0.03–1.91
MELD (5 point interval) 0.97 0.91–1.03 0.99 0.92–1.06 1.45 1.31–1.62 1.41 1.26–1.58 0.94 0.84–1.06 0.93 0.83–1.05 1.2 1.03–1.40 1.20 1.02–1.41
Open in a new tab

When including an interaction between in-state residency status and insurance status, in addition to adjustment for other candidate characteristics, ALD candidates (i.e., both ELT and SLT candidates) residing in Maryland remained less likely to progress from referral to evaluation if they were insured with Medicaid versus privately insured (HR 0.68, 95% CI 0.52–0.91). Out-of-state candidates with Medicaid insurance were significantly less likely than in-state candidates with Medicaid insurance to progress from referral to evaluation, relative to those with private insurance (HR 0.22, 95% CI 0.07–0.70).

Progression from Evaluation to Listing – Estimation of Hazard Ratio via Cox Regression

The median time from evaluation to listing for all candidates was greater than the 100-day interval of administrative censoring. The time for 35% of ELT candidates to progress from evaluation to listing was 36 days (95% CI 18–52), and the time for 35% of SLT candidates to progress from evaluation to listing was 52 days (95% CI 33–70). When stratified by SVI, the time for 35% of candidates to progress from evaluation to listing was 29 days (95% CI 11–42), 37 days (95% CI 23–53), and 86 days (95% CI 49-undefined) for low, moderate, and high SVI candidates, respectively (Figure 1).

Figure 1.

Figure 1.

Figure 1.

Open in a new tab

Time to progress from evaluation to listing by SVI for (a) ELT candidates, and (b) SLT candidates.

In unadjusted analysis, ELT candidates were less likely to progress from evaluation to listing if they resided in census tracts with high SVI versus those with low SVI (HR 0.61, 95% CI 0.39–0.96), if they had relevant psychiatric comorbidities (HR 0.37, 95% CI 0.17–0.79), or if they had inadequate social support (HR 0.37, 95%CI 0.17–0.80). After adjusting for other candidate characteristics, the difference by SVI was no longer statistically significant, but the effect of psychiatric comorbidities and social support remained significant (HR 0.42, 95% CI 0.19–0.94, and 0.44, 95% CI 0.20–0.95, respectively). There was no evidence of a significant association between age, sex, race, type of insurance, or distance to transplant center in progression from evaluation to listing.

In unadjusted analysis of the SLT cohort, candidates were less likely to progress from evaluation to listing if they resided in census tracts with moderate SVI versus those with low SVI (HR 0.54, 95% CI 0.29–0.99) or if they had relevant psychiatric comorbidities (HR 0.35, 95% CI 0.14–0.88). Candidates were more likely to progress from evaluation to listing if they were older (HR 1.40, 95% CI 1.06–1.83) or lived farther from the transplant center (HR 1.15, 95% CI 1.06–1.25). After adjusting for other candidate characteristics, the differences by age and SVI were no longer significant, but the differences by distance to transplant center and psychiatric comorbidities remained significant (HR 1.13, 95% CI 1.05–1.23, and HR 0.36, 95% CI 0.14–0.94, respectively). There was no evidence of a significant association between sex, race, type of insurance, or social support and progression from evaluation to listing (Table 3).

When including an interaction between ELT/SLT status and SVI, in addition to adjustment for other candidate characteristics, high SVI ELT candidates were no less likely than high SVI SLT candidates to progress from evaluation to listing, relative to low SVI candidates (HR 0.95, 95% CI 0.47–1.94).

Geographic-Level Analysis of Census Tracts with Referred Candidates

There were 9153 census tracts within a 150Nmi radius of our institution, of which 507 census tracts had at least one referral for either ELT or SLT (Table 4). There were 219 census tracts excluded for missing SDoH data. The majority of referrals came from census tracts that were predominantly white (72%) and within 50 Nmi of our institution (73%). The plurality of ALD referrals came from census tracts with low social vulnerability (44%). There was no difference in ALD mortality rates between census tracts with referrals and those without, with rates of 3.39 per 100,000 and 3.33 per 100,000, respectively (p = 0.16). There was a slightly lower percentage of individuals with AUD between census tracts with referrals and those without, at 5.51% and 5.59%, respectively (p <0.001). There was a slightly lower percentage reporting excessive drinking in census tracts with referrals compared to those without; these percentages were 17.9% and 18.7%, respectively (p < 0.001) (Table 4).

Table 4.

Selected demographic and socioeconomic characteristics of census tracts within 150Nmi of our institution. Individual-level characteristics are stratified by whether they reside within this radius.

Census Tracts
With ALD Referrals Without ALD Referrals p-value
 
Total (n) 507 8646
Racial Category a , n (%) 0.001
 Predominantly White 367 (72) 5811 (67)
 Predominantly Black 49 (10) 1052 (12)
 Predominantly Hispanic 0 (0.0) 199 (2)
 Racially Diverse 91 (18) 1584 (18)
Distance from Transplant Center, n (%) <0.001
 <50Nmi 371 (73) 1937 (22)
 50–100Nmi 102 (20) 2588 (30)
 100–150Nmi 34 (7) 4121 (48)
Overall SVI, n (%) <0.001
 Low Vulnerability 221 (44) 2740 (32)
 Moderate Vulnerability 182 (36) 2959 (34)
 High Vulnerability 104 (21) 2947 (34)
SVI – Socioeconomic status, n (%) <0.001
 Low Vulnerability 213 (42) 2768 (32)
 Moderate Vulnerability 193 (38) 2933 (34)
 High Vulnerability 101 (20) 2945 (34)
SVI – Household, n (%) <0.001
 Low Vulnerability 174 (34) 2805 (32)
 Moderate Vulnerability 206 (41) 2919 (34)
 High Vulnerability 127 (25) 2922 (34)
SVI – Minority status, n (%) 0.01
 Low Vulnerability 186 (37) 2828 (33)
 Moderate Vulnerability 186 (37) 2927 (34)
 High Vulnerability 135 (27) 2891 (33)
SVI – Transportation, n (%) <0.001
 Low Vulnerability 228 (45) 2736 (32)
 Moderate Vulnerability 160 (32) 2980 (35)
 High Vulnerability 119 (24) 2930 (34)
Estimated ALD Mortality Rates, per 100,000, Mean (SD) 3.39 (1.15) 3.33 (0.89) 0.16
Estimated AUD prevalence, Percent (SD) 5.51 (0.41) 5.59 (0.46) <0.001
Estimated Excess alcohol use prevalence, Percent (SD) 17.9 (1.21) 18.7 (1.11) <0.001
Open in a new tab
a

Predominant racial category was designated if >60% of the census tract was of a particular race. Racially diverse indicated that no race was greater than 60% of the total population.

Figure 2a shows that ALD referrals are represented in counties throughout the referral region, with no obvious correlation between ALD mortality rates (figure 2b). There are examples of fewer ALD referrals over census tracts with high SVI near our institution (the referral destination) to the west and east (Figure 2c and 2d).

Figure 2.

Figure 2.

Open in a new tab

Annual rate of ALD referrals by census tract (a, c). Bivariate choropleth map showing both overall composite SVI index by census tracts and relative ALD mortality rates (b, d). Panels a & b show the broader referral region, and c & d magnify the area directly adjacent to the transplant center, which has the highest concentration of referrals.

Rate of ALD referrals are expressed as the number of ALD referrals per 100,000 residents over age 18 per year, categorized into quartiles. SVI score tertiles correspond to 0 – 26, 27 – 62, and 63 – 100 of the national percentile rankings. Estimated ALD Rate tertiles correspond to 0 – 2.9, 2.9 – 3.8, and greater than 3.8 deaths per 100,000.

The final adjusted model included the number of ALD referrals per census tract, census tract population over age 18, census tract SVI, distance to transplant center, and estimated ALD mortality rate. Exploratory analyses found no difference in model effects when evaluating either the number of ELT candidates or the number of SLT candidates, prompting the decision to use the aggregated number of ALD candidates. Prevalence of AUD rates and excessive drinking were noncontributory and were excluded.

To understand which components of the SVI were associated with the outcome, separate models were performed. One model was performed for the composite SVI index, one model for each of the four components of the SVI, and another model with all four SVI components (Table 5).

Table 5.

Incidence rate ratios for number of ALD referrals by (a) overall SVI, and (b) each of the four SVI components.

A)
Incidence Rate Ratio (95% CI)
SVI (ref = low)
 Moderate Vulnerability 0.84 0.69–1.02
 High Vulnerability 0.53 0.42–0.66
Distance to transplant center (per 10 Nmi) 0.97 0.97–0.97
Estimated ALD Mortality rate (per death per 100k) 1.32 1.21–1.43
B)
SVI – SES Model SVI – Housing Model SVI – Race Model SVI – Transportation Model SVI - Fully Adjusted Model
IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI)
SVI – SES (ref = low vulnerability)
 Moderate vulnerability 0.99 0.82–1.20 -- -- -- -- -- -- 1.14 0.92–1.40
 High vulnerability 0.56 0.44–0.71 -- -- -- -- -- -- 0.81 0.59–1.10
SVI – Household
 Moderate Vulnerability -- -- 1.26 1.03–1.54 -- -- -- -- 1.35 1.10–1.66
 High Vulnerability -- -- 0.82 0.66–1.03 -- -- -- -- 1.12 0.87–1.44
SVI – Race/Ethnicity
 Moderate Vulnerability -- -- -- -- 0.73 0.59–0.89 -- -- 0.77 0.63–0.95
 High Vulnerability -- -- -- -- 0.50 0.40–0.62 -- -- 0.59 0.45–0.76
SVI – Transportation
 Moderate Vulnerability -- -- -- -- -- -- 0.70 0.58–0.86 0.77 0.63–0.95
 High Vulnerability -- -- -- -- -- -- 0.60 0.48–0.75 0.76 0.59–0.96
Distance to transplant center (per 10 Nmi) 0.73 0.71–0.74 0.73 0.71–0.75 0.71 0.69–0.73 0.73 0.71–0.75 0.97 0.96–0.97
Estimated ALD Mortality rate (per death per 100k) 1.32 1.22–1.44 1.29 1.19–1.40 1.25 1.15–1.35 1.31 1.21–1.43 1.30 1.19–1.42
Open in a new tab

After adjusting for distance to center and ALD mortality, high SVI was associated a decreased incidence of ALD referrals within a census tract compared to low SVI (IRR 0.53, 95% CI 0.42–0.66) (Table 5a). The Moran’s I value for the regression residuals was significantly greater than 0 (Moran’s I = 0.15, p < 0.001), suggesting the presence of residual spatial correlation. Efforts taken to account for this, such as including additional covariates and a spatial lag term, did not further reduce the spatial correlation of the models’ residuals. Given the residual spatial correlation was small, it was considered to not meaningfully influence the results.

Greater vulnerability for three of the four individual components of the SVI (socioeceonomic status, race, and transportation) were associated with fewer referrals for ALD (Table 5b). Moderate vulnerability in the household component of the SVI was associated with increased referrals for ALD when compared to low vulnerability (IRR 1.26, 95% CI 1.03–1.54), though this trend was not apparent for those residing in areas of high household vulnerability. When the four components of SVI were included together in the model in place of the overall composite index, the household,race/ethnicity, and transportation components had a significant association with the incidence of ALD referrals. There was a higher incidence of ALD referrals from census tracts with moderate household vulnerability compared to low household vulnerability (IRR 1.35, 95% CI 1.10–1.66, Table 5b). A stepwise decrease in the incidence of ALD referrals within a census tract was associated with both moderate and high prevalence of racial and ethnic minorities relative to low prevalence (IRR 0.77, 95%CI 0.63–0.95 and IRR 0.59, 95%CI 0.45–0.76, respectively), and a moderate and high prevalence of transportation vulnerability, relative to low prevalence (IRR 0.77, 95%CI 0.63–0.95 and IRR 0.76, 95%CI 0.59–0.96, respectively). (Table 5b).

Discussion

In this single-center study of 501 individuals referred for ELT, we followed candidates from referral to their endpoint and found that individuals with Medicaid insurance referred for ELT, but not SLT, were less likely to progress from referral to evaluation (Tables 2–3). After adjusting for psychosocial comorbidities and other candidate characteristics, ELT candidates from census tracts with high SVI were less likely to progress from evaluation to listing. This effect was significant on one of two regression analyses, and the magnitude of the effect was similar in both. This finding was not present for SLT candidates (Tables 2–3). When characterizing census tracts that were intended to be served by our institution, high SVI census tracts were less likely to have individuals referred for ALD compared to low SVI census tracts (IRR 0.53, 95% CI 0.42–0.66) after adjusting for levels of alcohol-related mortality.

We found that ELT candidates with Medicaid insurance were less likely to progress from referral to evaluation, similar to prior studies.15 Some insurers, including some Medicaid plans, restrict coverage for transplant to individuals with greater than 6 months of sobriety.45 This may explain why ELT candidates with Medicaid faced greater coverage restrictions than SLT candidates. Out-of-state candidates with Medicaid faced additional barriers to receiving access to transplant at our institution, relative to in-state candidates with Medicaid. Although Medicaid policy allows beneficiaries to receive care out of state when there is no in-state alternative,46 inter-state restrictions on Medicaid coverage may limit access to ELT for Medicaid patients.

We found that ELT candidates from census tracts with high SVI may be less likely to progress from evaluation to listing. Although this finding was not significant for SLT candidates in either regression, the effect estimates of SVI were similar for both ELT and SLT candidates, and the SLT population was limited by a smaller sample size. Alternatively, it’s possible this observation may result from a highly selective nature of ELT that is more challenging for particular groups than SLT. This finding of candidates from lower socioeconomic status being less likely to undergo evaluation and progress to listing is consistent with prior research16,47,25 and others showing the importance that SDoH have for transplant outcomes.18,13 The neighborhood in which an individual lives impacts several SDoH variables that may be important for transplant evaluation. Post-transplant recovery is a demanding process for patients and their social support systems, with intensive follow-up care and oftentimes hospital readmission. Listing requires a determination of adequate social support; however, there is ambiguity in this assessment and certain populations may find it harder to meet these criteria. For example, individuals from low-income neighborhoods may face a systematic disadvantage if they are perceived to have inadequate resource reserves to support themselves during post-transplant recovery. In addition, national metrics incentivize centers to maximize one-year survival post-transplant; however, there is no metric that incentivizes equity in access to transplant.48 These forces may shape access to this life-saving treatment.34,35

These neighborhood-level factors may also be linked to the ability of individuals to access the health system and be referred. We found that census tracts with high vulnerability had a lower rate of ALD referrals than census tracts with low vulnerability. We found that increasing vulnerability on three of the four components of the SVI were associated with a lower rate of ALD referrals. Greater racial and ethnic minority composition and decreased access to transportation were uniquely affected. Community factors such as access to alcohol treatment, access to gastroenterologists or transplant hepatologists, implicit bias among referring physicians, and insurance coverage may affect whether a referral is initiated for a particular patient.20,49,19 These factors may have a particularly prominent effect in neighborhoods with predominantly minority populations.

In our study, approximately two-thirds of candidates were referred during the COVID-19 pandemic. Prior research within transplant and other disciplines highlighted how the pandemic exacerbated disparities by race and SDoH.50,51 We were unable to determine to what extent the pandemic contributed to our observed association between social vulnerability and progression in transplant candidacy evaluation.

At our center, we have done the following to understand and mitigate the impact of SDoH on access to ELT. Our assessment of social needs for candidates at the point of evaluation is being further developed and standardized.52 These assessments have the potential for answering important research questions regarding how specific SDoH act as barriers in access to transplant, as well as supplementing clinical use when used in the appropriate context. We are also identifying a representative population with ALD in need of liver transplant to better characterize the underlying population in need of ELT. Identifying disparities between this population and the population ultimately transplanted may provide more precise insights than a geographic approach. We have plans to engage with experts familiar with our population prior to the point of referral and after transplant to provide valuable insights into who ultimately reaches our center and progresses through transplant evaluation that cannot be observed within the walls of our institution.

This study has several limitations. The retrospective design does not capture demographic and clinical characteristics for many individuals referred to but not seen at our center. An advantage is that our center prospectively collects data on individuals referred for liver transplant; however, it is often not feasible to collect complete data on patients not treated at our center. Additionally, these data do not include a comprehensive assessment of SDoH. The individuals excluded from our analysis were more frequently younger, of a racial minority, insured by Medicaid, and lived in census tracts with higher SVI. If complete data for all referred candidates was available, these sociodemographic factors may have shown a greater disadvantage in progressing from referral to evaluation than observed in our data. There are also likely additional individual-level SDoH that impact access and progression through evaluation to listing that are not captured within our referral registry. Although our center performs the largest number of early liver transplants in the area, we could not assess whether candidates with ALD needing ELT were referred to other centers. We also could not determine whether individuals needing ELT were representative of the census tracts in which they reside.

In conclusion, in a retrospective cohort of ALD candidates followed from the time of referral to the time of listing for transplant, insurance status and SVI are important factors at the stages of evaluation that ultimately influence who is eligible for transplant. To prevent SDoH from being a barrier to accessing ELT, transplant centers must understand the processes driving these findings. Identifying the factors that drive inequities in access is critical to designing appropriate interventions. Additionally, incentivizing transplant centers based on equity metrics rather than solely post-transplant outcomes may potentially accelerate these efforts. Ensuring candidates have fair opportunities in access to referring providers, as well as an opportunity for referral, and supporting candidates throughout transplant evaluation and post-transplant recovery, are integral components of creating an equitable system through which ELT and other forms of care for transplant populations can be provided.

Supplementary Material

1

Presentations:

  • James Flanary, MD, Po-Hung Chen, MD, Marie Diener-West, PhD, Michael R. Desjardins, PhD, MA, Elizabeth A. King, MD, PhD and Andrew M.. Cameron, MD. “Social Vulnerability And Referrals For Early Liver Transplant For Alcohol Liver Disease”. American College of Surgeons Clinical Congress 2023, Boston, MA. Oct 23, 2023.

    (This presented preliminary data of the geographic-level analysis of census tracts with referred candidates).

  • James Flanary, Kieranjeet Nijhar, Jonathan Mitchell, Kayleigh Herrick-Reynolds, Mayan Teles, Daniel Warren, Elizabeth King, Andrew Cameron. “Sociodemographic Characteristics of Candidates from Referral through Evaluation for Early Liver Transplant for Severe Alcoholic Hepatitis.” ASTS 23rd Annual Winter Symposium, Jan 12, 2023, Miami, FL.

    (This presented preliminary data of the individual-level analysis comparing stages of evaluation).

Acknowledgements:

The contents of this manuscript are the sole responsibility of the authors and do not necessarily reflect the views, assertions, opinions, or policies of the Department of Defense, or the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. government.

Financial support and sponsorship:

This work was funded in part by grants K23AA028297 (PI: Dr. Chen) and P50AA027054 (PI: Dr. Cameron) from the National Institutes of Health, and K08DK133638 (PI: Dr. Strauss) from the NIDDK. The content is solely the authors’ responsibility and does not necessarily represent the official views of the National Institutes of Health.

Abbreviations:

ALD

alcohol-associated liver disease

AUD

alcohol use disorder

ELT

Early liver transplantation

EMR

electronic medical record

SLT

Standard liver transplantation

Nmi

nautical miles

SAH

severe alcohol-associated hepatitis

SDoH

social determinants of health

SVI

Social Vulnerability Index

SW

social work

Footnotes

Disclosures: Po-Hung Chen eceived grants from Gilead Sciences Research Scholars Program. Ahmet Gurakar received grants from DURECT.

Assistance with the study: none

Data availability statement:

Clinical data that support the findings of this study are available on request from the corresponding author, AC. The data are not publicly available to protect privacy of research participants. Many of the data sources that support findings of this study are publicly available, including the social vulnerability index. (https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html), CDC Wonder (https://wonder.cdc.gov), Substance Abuse and Mental Health Services Administration (https://www.samhsa.gov/data/nsduh/2016-2018-substate-reports), County Health Rankings (https://www.countyhealthrankings.org/explore-health-rankings/rankings-data-documentation), US Census (https://data.census.gov), and US Census cartographic boundaries (https://www.census.gov/geographies/mapping-files/time-series/geo/cartographic-boundary.html).

References

  • 1.Lim JK, Keeffe EB. Liver transplantation for alcoholic liver disease: Current concepts and length of sobriety. Liver Transplantation. 2004;10(10):S31–S38. doi: 10.1002/jlts.2004.20267. [DOI] [PubMed] [Google Scholar]
  • 2.Lim J, Curry MP, Sundaram V. Risk factors and outcomes associated with alcohol relapse after liver transplantation. World Journal of Hepatology. 2017;9(17):771–780. doi: 10.4254/wjh.v9.i17.771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Chuncharunee L, Yamashiki N, Thakkinstian A, Sobhonslidsuk A. Alcohol relapse and its predictors after liver transplantation for alcoholic liver disease: A systematic review and meta-analysis. BMC Gastroenterology. 2019;19(1):150. doi: 10.1186/s12876-019-1050-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Singal AK, Mathurin P. Diagnosis and treatment of alcohol-associated liver disease: A review. J American Medical Association. 2021;326(2):165–176. doi: 10.1001/jama.2021.7683. [DOI] [PubMed] [Google Scholar]
  • 5.Lucey MR, Mathurin P, Morgan TR. Alcoholic hepatitis. New Engl J Med 2009;360(26):2758–2769. doi: 10.1056/NEJMra0805786. [DOI] [PubMed] [Google Scholar]
  • 6.Lim N, Kwong AJ, Jafri S, et al. Heterogeneity in center practices in liver transplantation for alcohol-associated liver disease in the united states. Am J Gastroenterol 2022;117(9):1530. doi: 10.14309/ajg.0000000000001863. [DOI] [PubMed] [Google Scholar]
  • 7.Mathurin P, Moreno C, Samuel, Dumortier J, Salleron J, Durand F, et al. Early liver transplantation for severe alcoholic hepatitis. New Engl J Med 2011;365(19):1790–1800. [DOI] [PubMed] [Google Scholar]
  • 8.Weeks SR, Sun Z, McCaul ME, Zhu H, Anders RA, Philosophe B, et al. Liver transplantation for severe alcoholic hepatitis, updated lessons from the world’s largest series. J American College of Surgeons. 2018;226(4):549–557. doi: 10.1016/j.jamcollsurg.2017.12.044. [DOI] [PubMed] [Google Scholar]
  • 9.Herrick-Reynolds KM, Punchhi G, Greenberg RS, Strauss AT, Boyarsky BJ, Weeks SR, et al. Evaluation of early vs standard liver transplant for alcohol-associated liver disease. JAMA Surgery. 2021;156(11):1026–1034. doi: 10.1001/jamasurg.2021.3748. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Maddur H, Asrani SK. Alcohol-associated hepatitis and liver transplantation: Mind the (racial, sex, economic, geographic, center, waitlist, and posttransplant outcomes) gap. Am J Transplantation. 2021;21(3):921–922. doi: 10.1111/ajt.16243. [DOI] [PubMed] [Google Scholar]
  • 11.Jakhete N, Abutaleb A, Shetty K. Transplant in acute alcoholic hepatitis: A relative contraindication. Current Opinion in Organ Transplantation. 2022;27(2):93–97. doi: 10.1097/MOT.0000000000000974. [DOI] [PubMed] [Google Scholar]
  • 12.Andreoni K, Zarrinpar A. The sobering complexities of alcoholic liver disease and decisions for transplant. JAMA Surgery. 2021;156(11):1034–1035. doi: 10.1001/jamasurg.2021.3749. [DOI] [PubMed] [Google Scholar]
  • 13.Wahid NA, Rosenblatt R, Brown RS. A review of the current state of liver transplantation disparities. Liver Transplantation. 2021;27(3):434–443. doi: 10.1002/lt.25964. [DOI] [PubMed] [Google Scholar]
  • 14.Thornton RLJ, Glover CM, Cené CW, Glik DC, Henderson JA, Williams DR. Evaluating strategies for reducing health disparities by addressing the social determinants of health. Health Affairs. 2016;35(8):1416–1423. doi: 10.1377/hlthaff.2015.1357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Bryce CL, Angus DC, Arnold RM, Chang CC, Farrell MH, Manzarbeitia C, et al. Sociodemographic differences in early access to liver transplantation services. Am J Transplantation. 2009;9(9):2092–2101. doi: 10.1111/j.1600-6143.2009.02737.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Strauss AT, Moughames E, Jackson JW, Malinsky D, Segev DL, Hamilton JP, et al. Critical interactions between race and the highly granular area deprivation index in liver transplant evaluation. Clinical Transplantation. 2023:e14938. doi: 10.1111/ctr.14938. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Cannon RM, Goldberg DS, Sheikh SS, Anderson DJ, Pozo M, Rabbani U, et al. Regional social vulnerability is associated with geographic disparity in waitlist outcomes for patients with Non–Hepatocellular carcinoma model for end-stage liver disease exceptions in the united states. Annals of Surgery. 2023;279(5). doi: 10.1097/sla.0000000000006097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Yang J, Endo Y, Moazzam Z, Lima HA, Woldesenbet S, Alaimo L, et al. Travel distance and social vulnerability index: Impact on liver-related mortality among patients with end-stage liver disease. Clinical transplantation. 2023;37(9):e15001. doi: 10.1111/ctr.15001. [DOI] [PubMed] [Google Scholar]
  • 19.Loy V M, Rzepczynski A, Joyce C, Bello S, Lu A . Disparity in transplant referral patterns for alcohol-related liver disease based on physician-dependent variables. Transplantation Proceedings. 2020;52(3):900–904. doi: 10.1016/j.transproceed.2020.01.015. [DOI] [PubMed] [Google Scholar]
  • 20.Warren C, Carpenter A, Neal D, Andreoni K, Sarosi G, Zarrinpar A. Racial disparity in liver transplantation listing. J American College of Surgeons. 2021;232(4):526–534. doi: 10.1016/j.jamcollsurg.2020.12.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.McElroy LM, Likhitsup A, Scott Winder G, Saeed N, Hassan A, Sonnenday CJ, et al. Gender disparities in patients with alcoholic liver disease evaluated for liver transplantation. Transplantation. 2020;104(2):293–298. doi: 10.1097/TP.0000000000002843. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Mathur AK, Ashby VB, Fuller DS, Zhang M, Merion RM, Leichtman A, Merion RM, Leichtman A, Kalbfleisch J. Variation in access to the liver transplant waiting list in the united states. Transplantation. 2014;98(1):94–99. doi: 10.1097/01.TP.0000443223.89831.85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Cigarroa F Invited commentary. J American College of Surgeons. 2021;232(4):534–535. doi: 10.1016/j.jamcollsurg.2020.12.033. [DOI] [PubMed] [Google Scholar]
  • 24.Yilma M, Cogan R, Shui AM, Neuhaus JM, Light C, Braun H, Braun H, Mehta N, Hirose R. Community-level social vulnerability and individual socioeconomic status on liver transplant referral outcome. Hepatology Communications. 2023;7(7):e00196. doi: 10.1097/hc9.0000000000000196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Mohamed KA, Ghabril M, Desai A, Orman E, Patidar KR, Holden J, et al. Neighborhood poverty is associated with failure to be waitlisted and death during liver transplantation evaluation. Liver Transplantation. 2022;28(9):1441. doi: 10.1002/lt.26473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Hughes RA, Heron J, Sterne JAC, Tilling K. Accounting for missing data in statistical analyses: Multiple imputation is not always the answer. International Journal of Epidemiology. 2019;48(4):1294. doi: 10.1093/ije/dyz032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.von Elm E Dr, Altman DG Prof, Egger M Prof, Pocock SJ Prof, Gøtzsche PC,MD, Vandenbroucke JP Prof. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. The Lancet. 2007;370(9596):1453–1457. doi: 10.1016/S0140-6736(07)61602-X. [DOI] [PubMed] [Google Scholar]
  • 28.ArcGIS pro (2023) [software]. version 2.8. ESRI. https://Www.esri.com/en-us/arcgis/products/arcgis-pro/overview. [Google Scholar]
  • 29.Underlying cause of death, 1999–2020 request. centers for disease control and prevention. https://wonder.cdc.gov. Accessed April 30, 2023.
  • 30.Peery AF, Crockett SD, Murphy CC, Jensen ET, Kim HP, Egberg MD, et al. Burden and cost of gastrointestinal, liver, and pancreatic diseases in the united states: Update 2021. Gastroenterology. 2022;162(2). doi: 10.1053/j.gastro.2021.10.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Golabi P, Paik JM, Eberly K, De Avila L, Alqahtani SA, Younossi ZM. Causes of death in patients with non-alcoholic fatty liver disease (NAFLD), alcoholic liver disease and chronic viral hepatitis B and C. Annals of Hepatology. 2021;27(1). doi: 10.1016/j.aohep.2021.100556. [DOI] [PubMed] [Google Scholar]
  • 32.Moon AM, Yang JY, Barritt AS, Bataller R, Peery AF. Rising mortality from alcohol-associated liver disease in the united states in the 21st century. Am J Gastroenterol 2019;115(1):79. doi: 10.14309/ajg.0000000000000442. [DOI] [PubMed] [Google Scholar]
  • 33.Kim D, Li AA, Gadiparthi C, Kahn MA, Cholankeril G, Glenn JS, et al. Changing trends in etiology-based annual mortality from chronic liver disease, from 2007 through 2016. Gastroenterology. 2018;155(4). doi: 10.1053/j.gastro.2018.07.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Substate estimates of substance use and mental illness from the 2016–2018 NSDUH: Results and detailed tables. substance abuse and mental health services administration https://www.samhsa.gov/data/nsduh/2016-2018-substate-reports.
  • 35.Flowerdew R, Green M, Kehris E. Using areal interpolation methods in geographic information systems. Papers in Regional Science. 1991;70(3):303–315. doi: 10.1111/j.1435-5597.1991.tb01734.x. [DOI] [Google Scholar]
  • 36.Centers for disease control and prevention/ agency for toxic substances and disease registry/ geospatial research, analysis, and services program. CDC/ATSDR social vulnerability index 2020 database US. [Google Scholar]
  • 37.Cogswell R, Masotti M, Morris AA, Hart A, Murray T, Yancy C. Assessment of U.S. heart transplantation equity as a function of race: Observational analyses of the OPTN database. Lancet Regional Health: Americas. 2022;13:100290. doi: 10.1016/j.lana.2022.100290. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Killian AC, Shelton B, MacLennan P, McLeod MC, Carter A, Reed R, et al. Evaluation of community-level vulnerability and racial disparities in living donor kidney transplant. Archives of surgery (Chicago. 1960). 2021;156(12):1120–1129. doi: 10.1001/jamasurg.2021.4410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.U.S. census bureau (2021). 2016–2020 American community survey 5-year public use microdata samples. https://data.census.gov. [Google Scholar]
  • 40.New national liver and intestinal organ transplant system in effect feb. April, 2020. https://unos.org/news/new-national-liver-and-intestinal-organ-transplant-system-in-effect-feb-4-2020/. Updated 2020. Accessed Apr 23, 2020.
  • 41.Zou G A modified poisson regression approach to prospective studies with binary data. Am J Epidemiology. 2004;159(7):702–706. doi: 10.1093/aje/kwh090. [DOI] [PubMed] [Google Scholar]
  • 42.R core team (2021). R: A language and environment for statistical computing. R foundation for statistical computing, vienna, austria. URL https://Www.R-project.org/. [Google Scholar]
  • 43.Moran PA. A test for the serial independence of residuals. Biometrika. 1950;37(1–2):178–181. doi: 10.1093/biomet/37.1-2.178. [DOI] [PubMed] [Google Scholar]
  • 44.Cavanaugh JE, Neath AA. The akaike information criterion: Background, derivation, properties, application, interpretation, and refinements. WIREs Computational Statistics. 2019;11(3):e1460. doi: 10.1002/wics.1460. [DOI] [Google Scholar]
  • 45.Lee BP, Vittinghoff E, Pletcher MJ, Dodge JL, Terrault NA. Medicaid policy and liver transplant for Alcohol-Associated liver disease. Hepatology (Baltimore, Md.). 2020;72(1):130–139. doi: 10.1002/hep.31027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Medicaid payment policy for out-of-state hospital services. . 2020. [Google Scholar]
  • 47.Henson JB, Chan NW, Wilder JM, Muir AJ, Mcelroy LM. Characterization of social determinants of health of a liver transplant referral population. Liver Transplantation. 2023;PublishAheadofPrint:0. doi: 10.1097/lvt.0000000000000127. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Strauss AT, Sidoti CN, Purnell TS, Sung HC, Jackson JW, Levin S, et al. Multi-center study of racial & ethnic inequities in liver transplant evaluation: Understanding mechanisms & identifying solutions. Liver Transplantation. 2022;00. doi: 10.1002/lt.26532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Vaeth PAC, Wang-Schweig M, Caetano R. Drinking, alcohol use disorder, and treatment access and utilization among U.S. racial/ethnic groups. Alcohol Clin Exp Res 2016;41(1). doi: 10.1111/acer.13285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Rosenblatt R, Lee H, Liapakis A, Lunsford KE, Scott A, Sharma P, et al. Equitable access to liver transplant: Bridging the gaps in the social determinants of health. Hepatology. 2021;74(5):2808. doi: 10.1002/hep.31986. [DOI] [PubMed] [Google Scholar]
  • 51.Macconmara M, Wang B, Patel MS, et al. Liver transplantation in the time of a pandemic. Annals of Surgery. 2021;274(3):427. doi: 10.1097/sla.0000000000004994. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Johnson KA, Barolin N, Ogbue C, Verlander K. Lessons from five years of the CMS accountable health communities model. Health Affairs Forefront. 2022. doi: 10.1377/forefront.20220805.764159. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2087699-supplement-1.pdf (342.2KB, pdf)

Data Availability Statement

Clinical data that support the findings of this study are available on request from the corresponding author, AC. The data are not publicly available to protect privacy of research participants. Many of the data sources that support findings of this study are publicly available, including the social vulnerability index. (https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html), CDC Wonder (https://wonder.cdc.gov), Substance Abuse and Mental Health Services Administration (https://www.samhsa.gov/data/nsduh/2016-2018-substate-reports), County Health Rankings (https://www.countyhealthrankings.org/explore-health-rankings/rankings-data-documentation), US Census (https://data.census.gov), and US Census cartographic boundaries (https://www.census.gov/geographies/mapping-files/time-series/geo/cartographic-boundary.html).

RESOURCES