Racial Disparities in Waitlist Outcomes of Patients Listed for Lung Transplantation

Abstract

BACKGROUND:

The Lung Allocation Score, implemented in 2005, prioritized lung transplant candidates by medical urgency rather than waitlist time and was expected to improve racial disparities in transplant allocation. We evaluated whether racial disparities in lung transplant persisted after 2005.

METHODS:

We identified all waitlisted adult lung transplant candidates in the United States from 2005–2021 using the Scientific Registry of Transplant Recipients. We evaluated the association between race and receipt of transplant using a multivariable competing risk regression model adjusted for demographics, socioeconomic status, lung allocation score, clinical measures, and time. We evaluated interactions between race and age, sex, socioeconomic status, and lung allocation score.

RESULTS:

We identified 33,158 candidates on the lung transplant waitlist between 2005–2021: 27,074 (82%) White, 3,350 (10%) African-American, and 2,734 (8%) Hispanic. White candidates were older, had higher education levels, and had lower lung allocation scores (p<0.001). After multivariable adjustment, African-American and Hispanic candidates were less likely to receive lung transplants than White candidates [adjusted subhazard ratio (95% Confidence Interval): (African-American) 0.86 (0.82, 0.91); (Hispanic) 0.82 (0.78, 0.87)]. Lung transplant was significantly less common among Hispanic candidates over age 65 (p=0.003) and non-White candidates from higher-poverty communities [(African-American) p=0.013, (Hispanic) p=.0036].

CONCLUSIONS:

Despite implementation of the Lung Allocation Score, racial disparities persisted for waitlisted African-American and Hispanic lung transplant candidates and differed by age and poverty status. Targeted interventions are needed to ensure equitable access to this life-saving intervention.

Graphical Abstract

Since 1963, the use of lung transplant (LT) to treat end-stage lung disease has increased dramatically.¹ In 2021, over 2,500 LTs were performed and demand for transplant continues to surpass supply.² As a result, equitable access to organs remains a key issue. The 2005 implementation of the Lung Allocation Score (LAS) aimed to maximize utility in transplant allocation by prioritizing LT candidates based on medical urgency and estimated therapeutic benefit rather than time on the waitlist.³ LAS had theoretical benefits for improving equity in waitlist priority, as there were racially biased delays in transplant listing in the prior system.^4–6

Since LAS implementation, reports of its impact on transplant equity have been mixed. In the early LAS period, one study reported elimination of racial disparities in transplant due to this allocation change.⁷ Others found an increase in non-White transplant recipients and an elimination of racial differences in survival after transplant.⁸ However, still others found that although non-White candidates presented at younger ages and with increased disease severity, they were persistently less likely to undergo LT after LAS.⁹ Although many studies have focused on racial disparities, these often coexist with inequities in other social determinants of health. For example, lower socioeconomic status is associated with adverse outcomes for patients receiving LT for idiopathic pulmonary fibrosis.¹⁰ Although previous studies have documented the existence of disparities in various social determinants of health including race and socioeconomic status, little is known about how these factors interact to impact LT outcomes.

As the LT community transitions to a new allocation system in 2023,¹¹ it is critical to understand and address inequities that existed during the LAS period. In this study, we evaluate whether receipt of LT differed by racial group among adult patients listed for LT between May 2005 and March 2021.

PATIENTS AND METHODS

DESIGN AND DATA SOURCES

This study was considered exempt by the Institutional Review Boards of New York University (i22-00146) and Johns Hopkins University (NA_00042871).

We conducted a retrospective cohort study. This study used data from the Scientific Registry of Transplant Recipients (SRTR). The SRTR data system includes data on all donors, waitlisted candidates, and transplant recipients in the US, submitted by the members of the Organ Procurement and Transplantation Network (OPTN). The Health Resources and Services Administration (HRSA), U.S. Department of Health and Human Services provides oversight to the activities of the OPTN and SRTR contractors. This dataset has been described elsewhere.¹²

STUDY POPULATION

We included adult patients listed for LT in the United States from May 2005 to March 2021. Follow-up data were available until August 2022. We excluded patients with prior LT or LAS scores below 20, which typically resulted from missing data. We included patients identified as Non-Hispanic White (White), African-American, or Hispanic due to small sample sizes for candidates of other races/ethnicities.

VARIABLES AND OUTCOMES

Our primary exposure variable was race. Additional demographic variables included age, sex, education level, employment status, and insurance type. We identified the patient’s geographic region and the proportion of residents below the federal poverty level in each patient’s ZIP code using data from the 2014 American Community Survey.¹³ This was used as a proxy measure of socioeconomic status (SES). We evaluated urban-rural status using Rural-Urban Commuting Area Codes.¹⁴ Clinical variables included extracorporeal membrane oxygenation (ECMO) use, ventilator dependence, body mass index (BMI), diabetes, predicted forced expiratory volume during the first second (FEV1), predicted forced vital capacity (FVC), LAS, primary diagnosis, and blood type. We also adjusted for year of LT. Values at the time of listing for transplant were used in this analysis.

Our primary outcome variable was receipt of LT. Waitlist mortality was a competing risk. Our secondary outcome was waitlist mortality; LT was a competing risk in these analyses.

STATISTICAL ANALYSIS

For categorical variables, we compared groups using percentages and Pearson’s Chi-square test. For continuous variables, we compared groups with means, standard deviations (SD) and ANOVA for normally distributed variables and medians, interquartile ranges (IQR) and Kruskal-Wallis tests for skewed data.

We estimated cumulative incidences using the Kaplan-Meier method and performed log-rank tests to compare unadjusted survival curves by race. We verified the proportional hazard assumptions by visually inspecting log-log plots and estimated adjusted subhazard ratios (aSHRs) using competing risks regression based on the Fine and Gray proportional subhazards model.¹⁵ Our adjusted model included variables likely to influence receipt of transplant, including age, sex, education level, insurance, blood type, diabetes status, ventilator use, BMI category, primary diagnosis, LAS, SES and year. Due to a priori hypotheses about effect modification, we assessed the relationships between race and other social determinants of health by evaluating whether associations between race and waitlist outcomes differed by age, gender, SES, and LAS score category (≤35, 35–45 and >45). Crude SHRs were estimated in models including race, the candidate variable and the interaction term. Adjusted SHRs were estimated by including the interaction term in the adjusted competing risks model. We performed all statistical analyses using Stata 16 (StataCorp LLC, College Station, TX). Two-sided p-values less than 0.05 were considered statistically significant.

MISSING DATA

Of the 33,158 records included in this analysis, 30,246 had complete data. We used multiple imputation to estimate missing values for educational status, insurance type, diabetes, and SES for 2,912 candidate records. Analysis of complete data was concordant with results from the imputed dataset (Supplemental Tables 3–7).

RESULTS

STUDY POPULATION

We identified 33,158 patients listed as LT candidates in the United States between May 2005 and March 2021. Of these, 82% (n = 27,074) of candidates were White, 10% (n = 3,350) were African-American, and 8% (n = 2,734) were Hispanic (Table 1). The mean age was 56 years (SD 12.7) and 43% of candidates were female.

Table 1:

Characteristics of Lung Transplant Candidates in National Registry (n=33,158) during 2005–2021 by Race.

Characteristics	Race Category
	Total (n=33,158)	White (n=27,074)	African-American (n=3,350)	Hispanic (n=2,734)	P value
Age at listing, mean (SD)	56.1 (12.7)	56.8 (12.8)	52.8 (11.1)	53.7 (13.2)	<0.001
Female, %	42.9	40.8	56.8	46.2	<0.001
Education, %					<0.001
Below high school	3.3	2.0	2.2	17.3
High School	39.6	39.0	40.6	43.6
Above high school	57.2	59.0	57.2	39.0
Life Support (ECMO) Use, %	1.9	1.7	2.4	3.1	<0.001
Life Support (Ventilator) Use, %	3.1	3.0	2.7	5.0	<0.001
Diabetes, %	18.5	17.6	19.9	25.9	<0.001
BMI, mean (SD)	25.7 (4.7)	25.6 (4.7)	26.4 (4.6)	26.1 (4.4)	<0.001
% Predicted FEV1, median (IQR)	34.0 (21.0, 52.0)	33.0 (21.0, 52.0)	37.0 (24.0, 51.0)	40.0 (28.0, 53.0)	<0.001
% Predicted FVC, median (IQR)	47.0 (37.0, 60.0)	48.0 (38.0, 61.0)	45.0 (34.0, 58.0)	41.0 (32.0, 52.0)	<0.001
LAS Score Category, %					<0.001
≤35	34.9	37.8	26.9	16.2
35–45	39.9	39.1	42.8	43.8
>45	25.2	23.1	30.3	40.0
Primary Diagnosis, %					<0.001
Obstructive Disease	28.8	31.8	21.1	8.8
Pulmonary Vascular Disease	4.9	4.2	7.9	7.7
Cystic Fibrosis/ Immunodeficiency Disorder	9.9	11.2	2.2	5.8
Restrictive Disease	56.3	52.6	68.7	77.2
Non-Specified	0.2	0.1	0.1	0.5
Other	0.0	0.0	0.1	0.0
Blood Type, %					<0.001
O	46.2	44.4	50.6	58.6
A	39.3	41.9	25.0	30.7
B	10.9	10.0	20.1	8.4
AB	3.6	3.7	4.3	2.4
Work for Income, %	15.0	15.6	13.1	12.0	<0.001
Insurance type, %					<0.001
Public	47.6	46.8	51.2	51.7
Private	51.7	52.5	48.4	47.9
None	0.6	0.7	0.4	0.4
Urban, %	96.9	96.4	99.4	99.0	<0.001
Region, %					<0.001
Northeast	19.8	20.0	20.2	17.3
Midwest	23.6	25.6	20.4	8.2
South	37.4	36.2	49.6	34.8
West	19.2	18.3	9.8	39.8
Poverty Rate (Percentage), median (IQR)	12.2 (7.4, 18.8)	11.3 (7.0, 17.4)	18.0 (11.0, 27.5)	17.2 (10.8, 24.7)	<0.001

BMI: body mass index; ECMO: extracorporeal membrane oxygenation, FEV1: forced expiratory volume during first second, FVC: forced vital capacity, SD: standard deviation; IQR: interquartile range.

PATIENT CHARACTERISTICS

White candidates were older (mean 56 years) than African-American (53 years) or Hispanic candidates (54 years, p < 0.001, Table 1). African-American candidates were more likely to be female (57%) than were Hispanic (46%) or White candidates (41%, p<0.001, Table 1). Hispanic candidates were more likely to have LAS>45 at listing (40%) than White (23%) or African-American candidates (30.3%, p<0.001, Table 1). Restrictive disease was more prevalent among African-American (69%) and Hispanic patients (77%) compared to White patients (53%, p<0.001). White patients had a higher prevalence of obstructive disease (32%) than their African-American (21%) or Hispanic counterparts (9%, p<0.001, Table 1). White candidates had higher levels of above high-school education. Hispanic candidates were more likely to have diabetes, O blood type, public insurance and higher FEV1, ECMO and ventilator support, lower FVC, and lower employment levels (p<0.001 for each comparison, Table 1). African-American patients had higher BMI and SES than other candidates (P<0.001, Table 1). Compared to White candidates, African-American candidates were significantly more likely to live in the South region and less likely to live in the West; Hispanic candidates were significantly more likely to live in the West and less likely to live in the Midwest (P<0.001, Table 1). White candidates were significantly less likely than African-American or Hispanic candidates to live in urban areas (P<0.001, Table 1).

RECEIPT OF TRANSPLANT

Of the 33,158 candidates, 27,246 (82%) underwent transplant. During the study period, 83% of White patients, 77% of African-American and 77% of Hispanic patients received LTs (p<0.001, Figure 1). Median time on the waitlist was 0.82 years for White patients, 1.09 years for African-American patients, and 0.87 years for Hispanic patients.

Figure 1.

Unadjusted Cumulative Incidence of Lung Transplant by Race Among Candidates in National Registry (N=33,158). Waitlist mortality was a competing event.

In an unadjusted competing risks analysis, compared to White candidates, African-American candidates (A) were 17% less likely to receive LTs (crude SHR 0.83, 95% CI 0.80–0.87) and Hispanic (H) candidates were 14% less likely than to receive LTs (crude SHR 0.86, 95% CI 0.82–0.90, Table 2). After adjustment, African-American candidates were 13% less likely to receive LTs than White candidates [adjusted SHR (aSHR) 0.87, 95% CI 0.83–0.91, Table 2]. Hispanic candidates were 17% less likely to receive LT than White candidates (aSHR 0.83, 95% CI 0.79–0.87, Table 2).

Table 2: Association between Race Category and Receipt of Lung Transplant.

Waitlist mortality was a competing risk. Crude (cSHR) and adjusted (aSHR) sub hazard ratios with 95% confidence intervals (95% CIs) derived from Fine and Gray competing risk models. Statistically significant associations (p<0.05) are bolded.

		cSHR (95% CI) n=33,158			aSHR* (95% CI) n=33,158
	White	African-American	Hispanic	White	African-American	Hispanic
Overall
	Ref	0.83 (0.80, 0.87)	0.86 (0.82, 0.90)	Ref	0.87 (0.83, 0.91)	0.83 (0.79, 0.87)
Gender
Male	Ref	0.90 (0.84, 0.95)	0.92 (0.86, 0.98)	Ref	0.90 (0.84, 0.96)	0.86 (0.81, 0.93)
Female	0.66 (0.64, 0.68)	0.58 (0.55, 0.61)	0.55 (0.52, 0.59)	0.69 (0.67, 0.71)	0.58 (0.55, 0.62)	0.54 (0.50, 0.58)
Age
<65 years	Ref	0.87 (0.84, 0.91)	0.90 (0.86, 0.95)	Ref	0.87 (0.83, 0.91)	0.87 (0.82, 0.92)
≥65 years	1.29 (1.25, 1.33)	1.06 (0.95, 1.19)	1.01 (0.91, 1.13)	1.15 (1.11, 1.19)	1.01 (0.90, 1.13)	0.83 (0.74, 0.92)
LAS
≤35	Ref	0.80 (0.75, 0.86)	0.77 (0.70, 0.85)	Ref	0.86 (0.80, 0.93)	0.84 (0.76, 0.92)
35–45	1.37 (1.33, 1.40)	1.06 (1.00, 1.12)	1.08 (1.02, 1.15)	1.35 (1.31, 1.39)	1.15 (1.08, 1.22)	1.13 (1.06, 1.21)
>45	1.57 (1.50, 1.64)	1.27 (1.16, 1.40)	1.22 (1.11, 1.33)	1.54 (1.47, 1.62)	1.39 (1.26, 1.53)	1.25 (1.14, 1.38)

^*Adjusted for age, gender, education, insurance, blood type, diabetes status, previous liver transplant, ventilator use and Body Mass Index, diagnosis, Lung Allocation Scores (LAS), poverty rate, year.

^¶Difference of difference (interaction)

INTERACTIONS

Racial differences in receipt of LT varied significantly by sex, age, and SES in our adjusted analysis (Table 3). Racial differences in LT did not vary by LAS score, urban location, or geographic region.

Table 3: Interactions in the Association between Race Category and Receipt of Lung Transplant.

Waitlist mortality was a competing risk. Crude (cSHR) and adjusted (aSHR) sub hazard ratios with 95% confidence intervals (95% CIs) derived from Fine and Gray competing risk models. Statistically significant associations (p<0.05) are bolded.

INTERACTIONS¶	cSHR (95% CI) n=33,158		aSHR (95% CI) n=33,158
	African-American	Hispanic	African-American	Hispanic
Sex
Female (vs. Male)	0.98 (0.91, 1.07)	0.92 (0.83, 1.00)	0.94 (0.86, 1.02)	0.91 (0.82, 0.998)
Age (years)
≥65 (vs. <65)	0.94 (0.84, 1.07)	0.87 (0.77, 0.98)	1.01 (0.89, 1.14)	0.83 (0.73, 0.94)
SES (per 5% poverty rate) ¶
	0.98 (0.96, 0.998)	0.98 (0.96, 1.01)	0.97 (0.96, 0.99)	0.97 (0.95, 0.998)
LAS (vs. <35)
35–45	0.96 (0.88, 1.05)	1.03 (0.92, 1.15)	0.99 (0.90, 1.08)	1.00 (0.89, 1.12)
>45	1.01 (0.90, 1.14)	1.01 (0.88, 1.15)	1.05 (0.93, 1.18)	0.97 (0.85, 1.12)

^*Adjusted for age, gender, education, insurance, blood type, diabetes status, previous liver transplant, ventilator use and Body Mass Index, diagnosis, Lung Allocation Scores (LAS), poverty rate, year.

^¶Difference of difference (interaction)

Compared to White men, Hispanic women were 46% less likely to undergo LT [aSHR 0.54 (0.50, 0.58), Table 2]. This accentuated, negative effect for Hispanic women is reflected in the statistically significant interaction between race and sex shown in Table 3 [aSHR 0.91 (0.82, 0.998)]. A similar, although not statistically significant, difference was observed in the relationship between race and LT by sex for African-American compared to White women [aSHR 0.94 (0.86, 1.02)].

For Hispanic (vs. White) candidates, the association between race and receipt of transplantation varied significantly by age [aSHR 0.83 (0.73, 0.94), Table 3]. Compared to younger White candidates, older Hispanic candidates were 17% less likely to receive LT [asHR 0.83 (0.74, 0.92), Table 2]. The association between race and receipt of transplant did not differ by age for African-American (vs. White) candidates.

Compared to White candidates, the association between race and receipt of LT varied significantly by SES for African-American [aSHR 0.97 (0.96, 0.99) for every 5% increase in SES] and Hispanic candidates [aSHR 0.97 (0.95, 0.998) for every 5% increase in SES, Table 3].

WAITLIST MORTALITY

African-American [crude SHR 1.29 (1.18, 1.41)] and Hispanic candidates [crude SHR 1.29 (1.17, 1.43)] were 29% more likely to die on the waitlist than White candidates in unadjusted analysis (Supplemental Table 1). After adjustment for covariates, there was no significant difference by racial group in waitlist mortality for African-American [aSHR 1.09 (0.99, 1.21)] or Hispanic candidates.

Compared to White candidates, the association between race and waitlist mortality varied significantly by age, and SES in our adjusted analysis (Supplemental Table 2). The association between race and waitlist mortality did not vary by sex, urban location, or geographic region. Among Hispanic (vs. White) candidates, the association between race and waitlist mortality differed significantly by age [aSHR 1.42 (1.12, 1.79)] and LAS [aSHR 0.78 (0.62, 0.97), Supplemental Table 2]. Older Hispanic patients were 44% more likely to die on the waitlist than younger White patients [aSHR 1.44 (1.17, 1.77), Supplemental Table 1]; by contrast, mortality was statistically no different for older vs. younger White patients [aSHR 1.04 (0.96, 1.13), Supplemental Table 1]. For African-American patients, waitlist mortality significantly differed by SES [aSHR 1.05 (1.00, 1.09), Supplemental Table 2].

COMMENT

In this large, national retrospective analysis of 33,158 LT candidates in the LAS era, African-American and Hispanic candidates were significantly less likely than their White counterparts to receive LT, even after adjusting for clinical and demographic factors. Racial differences in receipt of transplant were more pronounced for older Hispanic candidates, Hispanic women, and African-American and Hispanic candidates living in higher-poverty areas. Overall, adjusted waitlist mortality did not differ significantly by race. However, compared to White candidates, we found a stronger association between waitlist mortality and race for certain patient subgroups, including older Hispanic patients and African-American patients in higher-poverty areas.

Although by many measures, including LAS scores, ECMO use, and ventilator dependence, African-American and Hispanic candidates had higher clinical acuity at the time of listing, they were still less likely than White candidates to receive LTs in a system designed for allocation based on clinical need. Surprisingly, despite the higher acuity among African-American and Hispanic patients, racial disparities in LT did not vary by LAS. In addition, certain minority subgroups experienced higher rates of death on the waitlist, which may result from delays in receiving a transplant. This is even more profound since this analysis is limited to candidates on the transplant waitlist, who have already overcome the many barriers to accessing specialty care, passed evaluations of their insurance status and psychosocial support, and been deemed appropriate for LT.

Although there have been conflicting findings about racial disparities during the LAS era, our results are concordant with previous studies that have identified persistent disparities in LT. Using earlier SRTR data, Mooney and colleagues similarly identified racial disparities in receipt of transplant.⁹ We incorporate an additional 6 years of LT experience but demonstrate that racial disparities still persist. Riley et al reported a 23% increased odds of LT among White LT candidates compared to their non-White counterparts.¹⁶ We build upon these results by reporting outcomes for individual racial subgroups and using a more robust time-to-event analytic technique to address the competing risk of waitlist mortality.¹⁵ Our results also align with findings that for LT candidates with idiopathic pulmonary fibrosis, higher local income levels were associated with higher transplant rates.¹⁰ We expand upon this by including all patients listed for LT and evaluating the interaction between race and SES in influencing transplant outcomes. Our results also add to the growing body of evidence documenting sociodemographic and geographic disparities in access to LT and other forms of solid organ transplant. ^17–19 There has also been a recent focus on disparities in LT outside the United States; studies in France and Brazil have considered changes to their allocation systems due to sociodemographic and geographic disparities.^{20, 21}

Our results differ in some ways from the previous literature. Most notably, our results counter the findings of an early study demonstrating the apparent elimination of racial disparities in LT early after LAS implementation.⁷ This may be due to the longer observation period in our analysis and methodological differences which improve upon earlier work. Although previous studies of racial disparities only evaluated White and African-American or “non-White” candidates, we also include Hispanic candidates. Although Hispanic patients comprise a relatively small proportion of LT candidates, we demonstrate that they experience significant disparities that have received limited attention to date. Furthermore, previous studies have been limited in their evaluation of the interactions between race and other sociodemographic factors, which allows us to identify vulnerable groups of LT candidates for whom interventions to achieve transplant equity should be targeted.

This analysis has some limitations. First, our analysis was limited by the variables included in SRTR, which do not capture the full spectrum of social determinants of health, including individual financial data as well as information about local infrastructure or transportation. We used community-level poverty data from the American Community Survey as a proxy measure of socioeconomic status. Second, this analysis was limited to White, Hispanic, and African-American candidates due to low sample sizes for other racial/ethnic groups precluding multivariable analysis. Third, data for some demographic variables were missing and required multiple imputation. Our sensitivity analysis demonstrated that imputation did not meaningfully impact our results. Fourth, we did not directly compare the pre- and post-LAS implementation eras but instead analyzed the period of available data after LAS implementation and adjusted our analysis for the year of listing. Fifth, our analysis does not include removal from the waitlist for reasons other than death or transplant due to a high volume of missing data precluding reliable analysis. Sixth, the database used in this study includes data for the period between listing and transplant. Barriers to listing and post-transplant outcomes, including survival, may be subject to similar disparities but are outside the scope of this study.

In this large, national analysis, we demonstrate that despite hopes that organ allocation based on LAS would resolve inequities in LT, racial and socioeconomic disparities persisted in the LAS era and were more pronounced for several minority patient subgroups. Our study adds to a growing body of literature demonstrating disparities in LT, but little is known about the factors underlying these disparities or potential solutions. This may be due in part to the prior finding that racial disparities in waitlist outcomes were no longer present for lung transplant candidates in the LAS era. Despite the dearth of evidence in this area, recent studies suggest that barriers to lung transplant are likely systemic. For example, a recent study demonstrated that race-specific approaches to spirometry interpretation led to systematically lower calculated LAS for African-American patients, which may have contributed to racial disparities in lung transplant allocation.²² Additional barriers may include provider decision-making about real or perceived patient readiness, clinical stability, access to high-quality care while on the waitlist, and as yet unmeasured social determinants of health, including proximity to specialized care, social support, and economic factors, among others. Reasons for waitlist withdrawal and organ discard must also be carefully evaluated to identify processes that may exacerbate race-based disparities in access to lung transplant. In order to identify potential solutions, future studies will need to use qualitative and mixed-methods analytic techniques to evaluate patient-, provider-, and center-level barriers to equitable care in addition to challenges that can be addressed through broader policy change. Equity in lung transplantation will likely be achieved through a combination of improved allocation systems and targeted programs to support vulnerable transplant candidates, including racial minorities. As the United States transitions to a new LT allocation system in 2023, we must continually evaluate equity in LT to ensure that vulnerable transplant candidates, including racial minorities, are not left behind.

Abbreviations:

aSHR

Adjusted Subhazard Ratio

BMI

Body mass index

CI

95% confidence intervals

ECMO

Extracorporeal membrane oxygenation

FEV1

Predicted forced expiratory volume during the first second

FVC

Predicted forced vital capacity

LAS

Lung Allocation Score

LT

Lung Transplant

SES

Socioeconomic status (% below federal poverty line)

SHR

Subhazard ratios

SRTR

Scientific Registry of Transplant Recipients