Disparity in use of orthotopic liver transplantation among blacks and whites

Authors


Abstract

Orthotopic liver transplantation (OLT) is the best treatment for end-stage liver disease. Limited data exist on the access of minorities to OLT. The aim of this study was to determine whether disparities exist among black and white OLT patients. Data were collected from the United Network for Organ Sharing on black and white 18–70 year-old OLT waiting list registrants (n = 29,013) and OLT recipients (n = 15,805) between 1994 and 1998. Standardized transplant ratios were generated by comparing the racial distribution of OLT patients with the US population. Demographic and clinical characteristics of OLT registrants were compared by race. Multivariate analyses were performed to identify predictors of time to OLT and the likelihood of dying or receiving OLT within 4 years, controlling for severity of illness and other factors. The standardized transplant ratio for black OLT recipients (0.65) was significantly lower than the standardized transplant ratio for white OLT recipients (1.05). Blacks were younger and sicker than whites. After adjustment for severity and other factors, time to OLT among recipients did not differ by race (P > .05). Blacks were more likely to die or become too ill for OLT while waiting (P < .001). Blacks were less likely to receive OLT within 4 years (P < .001). In conclusion, adult blacks were underrepresented among OLT patients. Although waiting times were similar once listed, black race affected outcomes while awaiting OLT. The process of referral and evaluation for OLT should be investigated further. (Liver Transpl 2004;10:834–841.)

Orthotopic liver transplantation (OLT) is a life-saving procedure that is the treatment of choice for selected patients with end-stage liver disease (ESLD). The demand for cadaveric livers far outstrips the supply, and many patients die waiting for OLT. The United Network for Organ Sharing (UNOS) maintains the national organ transplantation waiting lists and develops policies for organ allocation. Organs are allocated according to priority of need based on severity of liver disease, geographic proximity to available organs, and until recently, waiting time. It is unclear whether this system results in equitable organ allocation among population groups.

Studies have shown that blacks have reduced access compared with whites for a number of medical therapies.16 The reasons for these disparities are multifactorial, involving medical, financial, and social factors, as well as patient and physician influences. Several studies have suggested that minorities may experience barriers to OLT,711 although the nature of such barriers has not been elucidated. This study had three objectives: (1) to determine whether blacks and whites are listed for and receive OLT at equal rates, (2) to determine whether the clinical characteristics at the time of OLT listing vary by race, and (3) to identify which factors influence the access to and timing of OLT after listing.

Abbreviations:

OLT, orthotopic liver transplantation; UNOS, United Network for Organ Sharing; ESLD, end-stage liver disease; MELD, Model for End-Stage Liver Disease; CI, confidence interval; HR, hazard ratio, OR, odds ratio.

Experimental Procedures

Date Sources and Study Population

The US Census Population Estimates for States by Age, Race, Sex, and Hispanic Origin: July 1, 1997 were used to establish black and white population estimates for adults between 18 and 70 years of age in the United States.12 The 1997 Census estimates included self-designated race (white; black; Asian or Pacific Islander; and Eskimo) and ethnicity (Hispanic or non-Hispanic). Black non-Hispanics and white non-Hispanics in each state were combined to generate black and white population estimates for the United States and for each UNOS region.

We examined UNOS data on all non-Hispanic black and non-Hispanic white patients between 18 and 70 years of age who were OLT waiting list registrants or OLT recipients between 1994 and 1998.13 Hispanics were not included because Hispanic ethnicity data is collected differently by OLT centers, often solely on the basis of surname, which renders the UNOS ethnicity data unreliable. UNOS variables evaluated included listing date, age, gender, blood type, listing diagnosis, UNOS region, primary insurance, waiting time (days), reason for removal from the OLT list, and 4 comorbid illnesses (systemic hypertension, angina/coronary artery disease, diabetes mellitus, and serum creatinine > 2.0 mg/dL). Listing diagnoses were grouped into the 8 most common diagnostic categories; all other diagnoses were combined into a category called “other.”

Status, a measure of severity of liver disease and urgency for OLT, also was obtained. Five status categories (1, 2, 3, 4, and 7) were used during the study period from January 1, 1994, to July 30, 1997. On July 30, 1997, the status designations were revised to 1, 2a, 2b, 3, and 7. Before July 30, 1997, a status 1 patient was in the intensive care unit because of acute or chronic liver failure, with a life expectancy without OLT of less than 7 days. After July 30, 1997, status 1 indicated acute hepatic failure (life expectancy less than 7 days). Before July 30, 1997, Status 2 indicated chronic liver failure requiring intensive care unit care or hospitalization. After July 30, 1997, status 2 was subdivided into status 2a and status 2b. Status 3 indicated active liver disease with Childs-Turcotte-Pugh score greater than 7. Status 4 was assigned to well-functioning patients with active liver disease managed at home. Status 7 was assigned to patients who became temporarily inactive on the OLT list.

Analysis

To determine whether blacks and whites received OLT at similar rates relative to the overall population, the racial distribution of the black and white OLT patients was compared with the racial distribution of 18–70 year-old black and white adults in the United States based on the 1997 US Census estimates. We calculated a standardized waiting list ratio and a standardized transplant ratio, which are similar in concept to standardized morbidity or mortality ratios.14 The standardized waiting list ratio is defined as the ratio of observed to expected number of patients on the OLT waiting list, whereas the standardized transplant ratio is the ratio of observed to expected number of OLT recipients. The expected number in both standardized ratios was obtained by applying the age distribution of black and white adults in the general population to the OLT cohorts. For each racial group, a standardized transplant ratio of 1 indicates that the number of patients receiving OLT was proportional to the distribution of that race in the general population. A standardized transplant ratio less than 1 indicates a lower number of OLT patients than expected, whereas a standardized transplant ratio greater than 1 indicates more transplants than expected based on population data. This analysis presupposes that the prevalence of ESLD and the need for OLT among blacks and whites parallels population data. ESLD prevalence data were not available; however, the age-adjusted US cirrhosis-associated mortality rate, defined as deaths per 100,000 population, is higher among blacks than whites.15 Given this mortality data, which suggest a greater prevalence of ESLD among blacks, standardizing the transplant ratio to the general population is likely a conservative measure. Thus, our estimate of the expected number of blacks in the transplant cohort may be conservative.

Next, we analyzed the UNOS data on the patients listed for or receiving OLT between 1994 and 1998. We compared by race the age, diagnosis, blood type, listing status, primary insurance, and comorbidities of black and white listed patients. Differences were tested using t tests for continuous variables and the chi-square tests for categorical variables. For the purpose of this analysis, patients with status 2, 2a, and 2b were combined into 1 category called status 2, then analyzed separately (2a and 2b) for patients on the OLT list after July 30, 1997.

We examined 2 measures of the likelihood of OLT after listing. First, we used the log rank test to examine the median time to OLT by race, age (younger than 50 years or 50 years old or older), gender, blood type, listing diagnosis, listing status, and primary insurer. Second, we used the chi-square test to compare the percentage of listed patients who received OLT within 4 years of listing. This analysis was restricted to those patients listed during calendar year 1994 to allow at least 4 years of follow-up (follow-up data were available through February 1999). Third, we compared the percentage of patients who died or became too sick for OLT within 4 years of listing. All univariate analyses were repeated after stratifying by status (1 vs. others).

We used several types of multiple regression models to estimate the likelihood of OLT, with control for confounders. A survival analysis of time to OLT was performed using a Cox proportional hazards model to estimate the simultaneous effects of race, age (in decades), gender, diagnosis, status, blood type, primary insurer, UNOS region, and comorbid illnesses. Those who did not receive OLT from 1994–1998 were censored. Cox proportional hazard models also were estimated for time to death or becoming too ill for OLT among those who had this outcome; all other patients were censored. Logistic regression analysis was performed to determine the predictors of OLT, and of death or becoming too sick for OLT, within 4 years of listing (restricted to patients listed in 1994). We repeated all multivariate analyses, stratifying patients into those listed before and after July 30, 1997. We also performed a stratified analysis by status (1 vs. others). For all multivariate models, we excluded patients with missing data, those who no longer needed OLT, as well as those who refused OLT or who were removed from the OLT list for unknown reasons (including those lost to follow-up).

All tests of significance were two sided. A P value of less than or equal to 0.05 indicated statistical significance. Results of the survival analyses and logistic regression analyses are reported as hazard ratios (HR) and odds ratios (OR), respectively. All statistical analyses were performed with the SAS statistical software package (Cary, NC).

Results

Table 1 shows the racial demographics of the black and white adult OLT waiting list registrants (n = 29,013), OLT recipients (n = 15, 805), and the black and white US population from 18 to 70 years of age. Blacks were underrepresented on the OLT waiting list (8.4%) and among OLT recipients (7.9%) compared with the general population, where blacks represented 13.6% of the total. For blacks, the standardized waiting list ratio and the standardized transplant ratio were 0.69 and 0.65, respectively, although the standardized waiting list ratio and standardized transplant ratio for whites was 1.04 and 1.05, respectively.

Table 1. Comparison of Racial Distribution of OLT Waiting List Registrants and OLT Recipients From 1994–1998 and US Racial Demographics from 1997
 Listed for OLTOLT RecipientsU.S. Population
N%Standardized Waiting List Ratio ± SE*N%Standardized Transplant Ratio ± SEUnadjusted Population Distribution (%)
  • Abbreviations: SE, standard error; OLT, orthotopic liver transplantation.

  • *

    Standardized transplant ratios calculated by dividing the number of OLT by the expected number of OLT based on the age distribution of the general population, age 18–70 years.

  • Unadjusted population distribution of blacks and whites 18–70 years old, based on the US Census estimates, 1997.

Black2,4368.40.69 ± 0.011,2557.90.65 ± 0.0213.6
White26,57791.61.04 ± 0.0114,55092.11.05 ± 0.0186.4
Total29,013100.0 15,805100.0  

Examination of UNOS OLT data from 1994 to 1998 showed that black patients were significantly younger and included more women than did the white patients (Table 2). The indication for OLT differed significantly by race. The distribution of blood types also varied by race; blood type B was more prevalent among blacks than whites. Significantly more blacks than whites had Medicaid, and more blacks than whites had hypertension and renal insufficiency. A significantly higher percentage of blacks was listed as status 1.

Table 2. Patient Characteristics of Black and White Adults on the OLT Waiting List, 1994–1998
 BlackWhiteP Values
  • *

    Other diagnoses include benign and malignant liver tumors, chronic hepatitis B virus infection, secondary biliary cirrhosis, Budd-Chiari Syndrome, metabolic liver diseases, cholangiocarcinoma, cirrhosis secondary to chemical exposure, and polycystic liver disease.

  • Status 2, 2a, and 2b were combined into status 2 for this analysis.

Patients listed for transplantationN = 2436N = 26,577 
Age (yrs)   
 Mean ± Standard deviation43.8 ± 12.049.4 ± 10.3< .001
Gender   
 Female53.0 %39.3 %< .001
Listing diagnosis  < .001
 Hepatitis C23.8 %24.7 % 
 Laënnec's (alcoholic) cirrhosis9.6 %17.1 % 
 Cryptogenic cirrhosis6.8 %10.5 % 
 Laënnec's cirrhosis and hepatitis C7.3 %7.3 % 
 Primary sclerosing cholangitis8.3 %7.0 % 
 Primary biliary cirrhosis2.5 %5.9 % 
 Autoimmune hepatitis9.2 %3.8 % 
 Acute hepatic necrosis7.9 %3.0 % 
 Other*15.9 %13.2 % 
 Not reported8.9 %7.5 % 
Listing Status  < .001
 114.7 %8.1 % 
 216.7 %12.3 % 
 356.7 %64.0 % 
 410.3 %14.1 % 
 71.4 %1.2 % 
 Not specified0.2 %0.3 % 
Blood type  < .001
 A24.8 %40.7 % 
 AB3.9 %3.8 % 
 B20.4 %10.7 % 
 O50.9 %44.7 % 
Insurance Payer  < .001
 Medicaid22.2 %11.1 % 
 Medicare10.7 %11.8 % 
 Other government3.9 %3.0 % 
 Private insurance52.1 %63.2 % 
 Self1.1 %1.4 % 
 Other10.0 %9.5 % 
Comorbidities   
 Diabetes mellitusN = 2167N = 23,751 
 14.3 %13.1 %.086
 Angina/coronary artery diseaseN = 2136N = 25,605 
 1.7 %2.7 %.003
 Serum creatinine > 2.0 mg/dl:N = 2154N = 23,662 
 10.1 %7.1 %< .001
 Drug-treated hypertension:N = 2107N = 23,234 
 14.3 %9.1 %< .001

Univariate analyses of the likelihood of OLT after listing revealed that the median time to OLT was significantly shorter for blacks than whites (282 days vs. 314 days, P = 0.022). Median waiting time also varied significantly according to age (younger patients had shorter waiting times), gender (men waited less time than women), diagnosis, listing status, blood group, primary insurer, renal function, and presence or absence of systemic hypertension (data not shown).

During 1994, 4,386 patients were listed for OLT; 392 patients were excluded from analysis because of missing information or clinical improvement, leaving 3,994 patients for analysis of 4-year outcomes. Overall, 73.5% underwent OLT within 4 years of listing. Fewer blacks (67.0%) than whites (74.2%) received OLT within 4 years (P = .003). The likelihood of OLT within 4 years also varied by age, diagnosis, blood type (lowest for type B), and listing status, with the lowest rate of OLT among status 1 patients. Within 4 years of listing, 25.6% of blacks and 17.8% of whites died or became too sick for OLT (P < .001). Rates of becoming too sick or dying before OLT also varied according to age, diagnosis, status, blood type, primary insurer, and renal function.

Results of the survival and logistic regression analyses are presented in Table 3. The time to OLT among recipients from 1994–1998 was similar for blacks and whites, after controlling for age, gender, listing diagnosis, status, geographic location, blood type, primary insurer, and comorbid illnesses (HR, 0.95, 95% confidence interval [CI], .90 to 1.01, P = .12). Among patients who did not receive OLT, black race was associated with a shorter time to death or becoming too sick before OLT compared with whites (HR, 1.36, 95% CI, 1.23 to 1.49, P < .001). The logistic regression analyses of patients listed in 1994 showed that blacks were significantly more likely to die or become too sick for OLT within 4 years of listing (OR 1.52, CI, 1.152 to 1.993, P = .003) and less likely to receive OLT within 4 years compared with whites, even after controlling for multiple confounding factors, including status (OR = 0.67, 95% CI, 0.52 to 0.87, P = .003). There was no significant interaction between race and gender or race and blood type (P > .05).

Table 3. Multivariable Analyses of Predictors of Time to OLT, Likelihood of OLT Within 4 Years of Listing for OLT, and Likelihood of Death Within 4 Years While Waiting for OLT
VariableTime to OLT*Likelihood of Transplant Within 4 YearsLikelihood of Death or Too Sick for OLT Within 4 Years
N = 26,720N = 3,994N = 3,994
Hazard RatioP ValueOdds RatioP ValueOdds RatioP Value
  • Abbreviations: HR, hazard ratio; OR, odds ratio; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis.

  • *

    Time to OLT (1994–1998) performed using Cox proportional hazards model. Patients who did not receive OLT were censored.

  • Regression analyses to determine predictors for OLT within 4 years of listing, and predictors of dying/becoming too sick for OLT within 4 years of listing restricted to patients listed in 1994. All multivariate models controlled for race, age, gender, diagnosis, status, blood type, primary insurer, geographic region, and comorbidities.

Race/ethnicity      
 White (reference level)
 Black0.95.1200.67.0031.52.003
Age (by increasing decade)0.96< .0010.90.0041.15< .001
Gender      
 Male (reference level)
 Female0.89< .0010.89.1401.08.420
Listing diagnosis      
 Hepatitis C0.79< .0010.70.0651.46.095
 Laënnec's (alcoholic) cirrhosis0.78< .0010.53.0011.66.026
 Cryptogenic cirrhosis0.83< .0010.59.0111.66.036
 Laënnec's cirrhosis and hepatitis C0.77< .0010.64.0761.86.033
 Primary sclerosing cholangitis0.91.0270.81.3501.45.15
 PBC (reference level)
 Autoimmune hepatitis0.76< .0010.48.0021.73.048
 Acute hepatic necrosis1.39< .0010.45.0032.26.005
 Other0.91.0160.46< .0012.34< .001
Listing status      
 15.96< .0010.52< .0012.49< .001
 22.48< .0010.83.151.60< .001
 3 (reference level)
 40.91< .0010.50< .0011.21.120
 71.12.1300.44.0041.94.033
Blood type      
 A (reference level)
 AB1.46< .0011.43.110.76.280
 B0.97.2200.67.0011.58< .001
 O0.72< .0010.73< .0011.22.040

We reanalyzed the data, stratifying patients into those listed before (n = 19,265) and those listed after (n = 9748) July 30, 1997. Before July 30, 1997, 16.1% of blacks and 9.3% of whites were listed as status 1; after the status change, 11.9% of blacks and 5.7% of whites were listed as status 1, and significantly more blacks than whites were listed as status 2a and 2b. Black race remained a significant predictor of dying or becoming too sick for OLT before July 30, 1997 (HR, 1.39, 95% CI 1.25 to 1.56, P < .001) and after July 30, 1997 (HR, 1.26, 95% CI 1.06 to 1.50, P = .008).

Finally, all analyses were repeated after stratification by status. In univariate analysis, status 1 black patients (n = 358) had longer waiting time to OLT compared with whites (n = 2147) (9 days vs. 7 days, P = .030). Time to OLT among transplanted patients of other statuses was not different between blacks (n = 2073) and whites (n = 24356). In 1994, 417 whites and 57 blacks were listed as status 1; 46% of blacks and 33% of whites died or became too sick while waiting for OLT (P = .057). Among those of other status, 22% of blacks and 16% of whites died or became too sick before OLT (P = .006). Table 4 shows the stratified multivariate analyses. In both strata, blacks had shorter time to death or becoming too sick for OLT compared with whites. Blacks listed as status 2–7 were significantly less likely to receive OLT and more likely to die within 4 years compared with whites. There was no racial difference in likelihood of receiving OLT among status 1 patients.

Table 4. Multivariate Analysis of Impact of Black Race Stratified by Status on OLT Outcomes
 Status 1 PatientsNon-Status 1 Patients
  • Abbreviations: HR, hazard ratio; OR, odds ratio; OLT, orthotopic liver transplantation; CI, confidence interval.

  • *

    Time to OLT for all transplanted patients, 1994–1998. Patients who did not receive OLT were censored.

  • Time to removal from OLT list because of death/too sick from 1994–1998. Patients who did not die or become too sick were censored.

  • Logistic regression restricted to patients listed in 1994.

  • Reference group for black race is white race.

1994–1998N = 2120N = 24,600
 HR95% CIP ValueHR95% CIP Value
Black race      
 Time to OLT*0.840.71, 1.00.0480.980.92, 1.04.460
 Time to death1.321.07, 1.63.0081.331.19, 1.48<.001
1994N = 474N = 3520
 OR95% CIP ValueOR95% CIP Value
Black race      
 OLT within 4 years0.720.38, 1.40.3400.720.54, 0.95.020
 Died/too sick within 4 years1.510.78, 2.92.2201.481.09, 2.02.013

Discussion

This study uses national data from 1994 to 1998 to examine the use of OLT among blacks and whites in the United States. This study shows that blacks were underrepresented on OLT lists, blacks and whites on the OLT list differed significantly with regard to medical and sociodemographic characteristics, blacks were more likely than whites to die while waiting for OLT, and blacks with chronic disease were less likely to receive OLT within 4 years compared with whites. These findings raise important questions about the access to and timing of OLT among blacks compared with whites.

Blacks, but not whites, were significantly underrepresented among OLT patients when compared with the racial distribution of the general population. Two previous studies, limited to 1–3 OLT centers, have suggested that adult blacks receive fewer liver transplants than expected.7, 8 Our study confirms and expands on these earlier studies by examining the entire OLT database over a 5-year period. Racial differences between a treatment population and the general population are subject to competing explanations, most importantly, racial differences in disease prevalence. No national data for ESLD prevalence exist, but there is indirect evidence that blacks carry the heaviest burden of ESLD with the worst outcomes. Blacks have higher mortality from cirrhosis than whites.15, 16 Blacks have higher prevalence of hepatitis C,17 are less likely than whites to receive or respond to antiviral treatment,18, 19 and have higher inpatient mortality from hepatitis C–related liver disease.20 Blacks have more alcoholic liver disease,21, 22 chronic hepatitis B infection,23 and hepatocellular carcinoma24 than whites. This racial disparity in apparent disease burden and OLT rates highlights the need to explore in greater detail the link between ESLD and access to treatment.

The low representation of blacks on OLT lists raises important questions about the OLT referral and evaluation process. Many variables influence the access of patients with ESLD to OLT, including patient preference, referral patterns, and appropriateness for OLT. The determination of appropriateness is based on objective and subjective criteria and is not necessarily proportional to the prevalence of ESLD. Differences in appropriateness for OLT may explain the racial differences in standardized transplant ratios. The referral process may represent a significant and disproportionate barrier to OLT for black patients with ESLD, as it does for black patients with end-stage renal disease (ESRD);1, 25 however, we were unable to evaluate the referral process. Our study does show that blacks had more severe liver disease than whites at the time of listing and were more likely to die while waiting, which likely reflects more advanced disease among black patients at the time of referral. Examination of the referral and evaluation process for OLT among different racial groups should be the focus of additional investigation.

Black patients with ESLD who were listed for OLT differed significantly from whites in age, severity of illness, and listing diagnoses. Outcomes for blacks on the OLT list were also worse than whites, even among the most acutely ill patients (status 1). After controlling for confounding factors, the waiting time among recipients of OLT was similar, but blacks died quicker and more often than whites while waiting. Consequently, blacks were less likely to receive OLT within 4 years. Our findings are not explained easily but are consistent with other studies that have shown that blacks encounter barriers to sophisticated medical therapies.16 In this study, as in others, race is likely a proxy for unmeasured factors, such as lower socioeconomic status, which may influence access to and quality of medical care.2628 Racial differences in comorbidities, education, and financial resources might affect access to and survival during the OLT process. The need to identify and quantify the effects of these factors on OLT is illustrated in our study as well as a recent study that showed an association between black race and poor outcome after OLT due, in part, to a higher number of gravely ill black patients.29 These studies support the need for well-designed, prospective, outcome studies of the pre-OLT and post-OLT processes among black patients with ESLD.

The strengths of our study include the size and detail of our database, which permitted analyses of multiple factors that may influence access to OLT. In addition, we used the best estimates of population statistics to determine standardized transplant ratios. There are several limitations as well. First, we used a large administrative database, and the integrity of our results depends on the accuracy of UNOS data. Second, we were unable to control for the site of care, which could influence access to OLT. Finally, we recognize that the racial disparities found in this retrospective study may not reflect current listing practices and outcomes now that the Model for End-Stage Liver Disease (MELD) system is used to determine severity of ESLD and priority for OLT.30 Preliminary data show that blacks have higher mean MELD scores at listing than whites,31 and a recently published study shows that the number of blacks removed from the OLT list because of death or becoming too sick has not been altered by the MELD system.32 It will be important to repeat these analyses with contemporary data to determine whether MELD will affect the disparity in outcomes between blacks and whites in need of OLT.

Most of the literature on access to transplant services has focused on ESRD. This work has identified biological and social impediments to renal transplantation for blacks, including low referral rates,1, 2, 33 severity of illness and comorbid disease,3335 low availability of compatible organs,36 socioeconomic factors,3335, 37 and patient preferences.1, 38 Little research has focused on the access of minority populations to OLT, yet these populations may face similar barriers. We have demonstrated significant differences between blacks and whites in the use of OLT, including differences in clinical characteristics, OLT rates, and survival while waiting for OLT. Our results highlight the need for further research to identify the barriers to OLT faced by black patients with ESLD and to identify those factors that are amenable to change.

Acknowledgements

The authors thank the United Network of Organ Sharing (UNOS) for the use of its data.

Ancillary

Advertisement