SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Having complications of portal hypertension is a harbinger of decompensated cirrhosis and warrants consideration for liver transplantation (LT). Racial disparities in LT have been reported. We sought to characterize disparities in the performing of surgical and endoscopic procedures among hospitalized patients with complications of portal hypertension. We queried the Nationwide Inpatient Sample from 1998 to 2003 to identify patients with cirrhosis and complications of portal hypertension. Logistic regression controlling for confounders was used to evaluate race as a predictor of undergoing a portosystemic shunt and LT and of dying in the hospital. Compared to whites, the adjusted odds ratios of receiving a portosystemic shunt were 0.37 (95% CI: 0.27-0.51) and 0.69 (95% CI: 0.54-0.88) for African Americans (AAs) and Hispanics, respectively. AAs with variceal bleeding were more likely to have endoscopic variceal hemostasis delayed more than 24 hours after admission than were whites (OR 1.6; 95% CI: 1.2-2.1). The adjusted odds ratios of undergoing LT were 0.32 (95% CI:0.20-0.52) and 0.46 (95% CI: 0.25-0.83) for AAs and Hispanics, respectively. Compared to whites, AAs experienced higher in-hospital mortality (OR 1.12; 95% CI: 1.01-1.24), whereas Hispanics had a lower risk of death (OR 0.83; 95% CI: 0.75-0.92). Among variceal bleeders, the odds ratio of death for AAs was 1.7 (95% CI: 1.2-2.4) compared to whites. Conclusion: AAs and Hispanics hospitalized for complications of portal hypertension were less likely to undergo a palliative shunt or LT than whites, which may contribute to the higher in-hospital mortality of AAs. Further studies are warranted to elucidate the mechanisms of these exploratory findings. (HEPATOLOGY 2007;45:1282–1289.)

End-stage liver disease or cirrhosis is the 12th leading cause of death in the United States.1 Patients with chronic liver disease develop portal hypertension–related complications at a rate of about 5%-7% per year, marking the transition from compensated to decompensated cirrhosis.2 Patients with cirrhosis frequently require hospitalization for complications of cirrhosis including variceal bleeding, ascites, and hepatic encephalopathy. In accordance with standard guidelines, these decompensated patients warrant consideration for liver transplantation, which is lifesaving and markedly improves the quality of life.3 Without transplantation, the 2-year survival for patients with decompensated cirrhosis is less than 50%.4 While awaiting transplantation, up to 30% of patients with decompensated cirrhosis become candidates for temporizing palliative procedures including endoscopic band ligation and portosystemic shunts.

In a seminal publication in 2002, the Institute of Medicine brought to light racial health disparities pervading treatment of the entire spectrum of chronic and subacute diseases.5 Several studies have suggested there may be racial barriers to liver transplantation (LT).6–10 However, the selection of these study populations may have been biased because patients were either from tertiary transplant centers or from the United Network for Organ Sharing (UNOS) database, a selected subgroup that has already been referred to and evaluated for LT. At a population level, issues of racial disparity may be even more pervasive. These findings raise pressing concerns, given that AAs carry a disproportionately high burden of hepatitis C,11 the leading cause of end-stage liver disease in the United States. Furthermore, the United States is in the midst of a demographic shift to minorities making up more than half the population by 2050.12 Thus, health disparities in the management of conditions as prevalent as cirrhosis will certainly affect the overall health of the nation.

In this study, we assessed the influence of race and health insurance on the likelihood of receiving portosystemic shunt procedures, delayed endoscopic variceal hemostasis, and liver transplantation in a large, nationally representative population-based sample of hospitalized patients with cirrhosis and complications of portal hypertension

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Data Source.

All data were extracted from the Nationwide Inpatient Sample (NIS) for the years 1998-2003. It is the largest all-payer database of national hospital discharges, maintained as part of the Healthcare Cost and Utilization Project (HCUP) by the Agency for Healthcare Research and Quality (AHRQ). The NIS is a 20% stratified sample of acute-care hospitals in the United States. It is a sample of data from community and general hospitals and academic medical centers but not long-term facilities. Each record in the NIS is for a single admission, not a person. Therefore, within a data set, there could be multiple records for one individual, if that individual had several admissions. Each data entry includes a unique identifier, demographic data (age, sex, and race), hospital transfer status, admission type (emergent, urgent, or elective), primary and secondary diagnoses (up to 15), primary and secondary procedures (up to 15), expected primary and secondary insurance payers, total hospital charges, length of stay, and hospital characteristics (region, location [urban or rural], size [by number of beds], teaching status). The NIS data concur with the National Hospital Discharge Survey, supporting the reliability of the NIS.13

Eligibility Criteria.

We included all hospital discharges between 1998 and 2003 with data on race that had the following as either a primary or secondary diagnosis: (1) cirrhosis (571.2, 571.5, 571.6) and (2) portal hypertension (572.3) or complications of portal hypertension including ascites (789.5), spontaneous bacterial peritonitis (567.23), hepatic encephalopathy (572.2), and variceal bleeding (456.0, 456.2). The above diagnoses and in-hospital procedures were identified by Clinical Modification of the International Classification of Diseases, 9th Revision (ICD-9-CM) diagnostic and procedural codes. Patients with cirrhosis who were electively admitted (identified by the variable admission type) in preparation for liver transplantation on the same day were excluded because of concern that they had a level of disease acuity different from that of those hospitalized for complications of portal hypertension.

Predictor and Outcome Variables.

Data on race and ethnicity were derived from hospital administrative data and included the following categories: white, AA, Hispanic, Asian or Pacific Islander, Native American, and “other.” In our analysis the latter 3 classifications were combined into an “other” category because of the small numbers in those classifications. Also in our analysis, we had ethnicity take precedence over race, and so Hispanic patients were coded as “Hispanic,” while all other race categories were non-Hispanic. Case-mix adjustment was performed using the well-validated Deyo modification of the Charlson Index (CI).14, 15 The Charlson Index is an instrument widely used to characterize and adjust for disease burden and case mix in administrative data. This scale incorporates 17 comorbid conditions that are weighted to yield a summary index. The summarized Charlson Index score is a validated predictor of all-cause in-hospital mortality16 but has not been validated specifically for liver-related disease. Surgical and endoscopic procedures were identified with these ICD9-CM procedural codes: 50.50, 50.51, and 50.59 for liver transplantation; 39.1 for portosystemic shunt; and 45.13, 44.13, 42.23, and 42.33 for upper endoscopy/variceal hemostasis. The validity of these codes for major surgical procedures has been documented.17 The preoperative interval was calculated from the difference between the day of surgery and admission date. For subjects who underwent surgery on the same day as admission, the preoperative interval was considered 0.5 days.

Statistical Analysis.

Data were analyzed using the Stata 9.0 SE software package (Stata Corp LP, College Station, TX). Analyses took into account a stratified 2-stage cluster design using Stata's SVY (survey data) commands, which incorporated discharge-level weights published by the HCUP. Two-way χ2 analyses were performed to compare discharge-level categorical variables among different racial groups. Logistic regression was used to calculate odds ratios for portosystemic shunt procedures, LTs, and in-hospital deaths. These analyses adjusted for age, sex, primary health insurance carrier, Charlson Index, calendar year, complications of portal hypertension (presence of ascites, variceal bleeding, encephalopathy, hepatorenal syndrome), and hospital characteristics (region, size by number of beds, rural versus urban location, and teaching status). These regression models were constructed by including potential confounders as covariates. Interaction between race and type of health insurance was also assessed in these models. Because complications of portal hypertension became less predictive of undergoing LT after incorporation of the model for end-stage liver disease (MELD) in 2002, a sensitivity analysis of admissions before 2002 was performed.18 A subgroup analysis of variceal bleeders was performed to assess racial differences in upper endoscopy being used for variceal hemostasis. Multiple logistic regression was similarly used to derive the influence of race and health insurance on the likelihood of having a delay in receiving variceal hemostasis (defined as more than 24 hours after admission).

Ethical Considerations.

The research protocol was approved by the Institutional Review Board of the Johns Hopkins Medical Institutions.

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

The baseline demographics of the 63,696 patients with cirrhosis who were admitted for a complication of portal hypertension are shown in Table 1. AAs and Hispanics were younger than whites (54.8 years and 57.2 years, respectively, versus 59.2 years, P < 0.001). AAs and Hispanics were also much less likely to have private insurance and more likely to receive Medicaid than whites. There were notable differences in the distribution of race by geographic region, with a large proportion of Hispanics and Asians residing in the West. AAs were also much more likely than whites to be admitted to a teaching hospital (62.4% versus 44.1%, P < 0.001). Minorities were less likely to have ascites and more likely to have encephalopathy (Table 2). However, there were no racial differences in having hepatorenal syndrome.

Table 1. Demographics of Patients with Cirrhosis in the NIS 1998-2003 by Race and Ethnicity
Demographic variableWhite (N = 41,633)African American (N = 6,962)Hispanic (N = 11,888)Asian (N = 1,211)Other (N = 2,002)
  • NOTE. “Other” indicates Native American or other race.

  • *

    P < 0.001, compared to white patients.

  • P < 0.01, compared to white patients.

  • P < 0.05, compared to white patients.

  • §

    The category “Large” was defined as being in the top tertile in size (number of beds) for each state.

Age (SE)59.2 (0.2)54.8 (0.2)*57.2 (0.4)*61.1 (0.7)55.2 (0.4)
Female16,267 (39.1)2,941 (42.2)4,198 (35.3)*513 (42.6)716 (35.6)
Health Insurance     
 Private11,113 (27.1)1,410 (20.7)a1,915 (16.4)*309 (25.7)*451 (22.8)*
 Medicare18,976 (46.2)2,446 (35.6)a4,516 (38.3)*458 (37.9)*569 (28.6)
 Medicaid7,035 (17.2)2,280 (33.4)*3,660 (31.4)*354 (29.5)*665 (34.2)
 Self-pay2,590 (6.2)490 (7.0)*1,053 (8.9)*53 (4.5)*164 (8.3)
 Other1,355 (3.3)226 (3.3)*582 (4.9)*30 (2.4)*124 (6.2)
Region     
 Northeast10,565 (26.5)1708 (26.2)*1,973 (17.9)*198 (17.4)*774 (40.9)
 Midwest5,579 (14.1)780 (11.6)*165 (1.5)*40 (3.4)*258 (13.3)
 South18,115 (42.0)3715 (51.6)*4,472 (36.5)*181 (14.3)*570 (26.6)
 West7,374 (17.4)759 (10.6)*5,278 (44.0)*792 (64.8)*400 (19.2)
Location     
 Rural5,547 (13.8)528 (7.8)*586 (5.2)*74 (5.9)161 (8.1)
 Urban36,069 (86.2)6,429 (92.2)*11,302 (94.8)*1,137 (94.1)1,841 (91.9)
Hospital Size§     
 Small or medium16,197 (38.2)2,815 (39.7)5,019 (41.3)460 (37.5)742 (36.2)
 Large25,419 (61.8)4,142 (60.3)6,869 (58.7)751 (62.5)1,260 (63.8)
Hospital Type     
 Nonteaching23,805 (55.9)2,700 (37.6)6,036 (49.2)603 (47.9)803 (38.0)
 Teaching17,811 (44.1)4,257 (62.4)*5,852 (50.8)608 (52.1)1,199 (62.0)*
Table 2. Clinical Presentation of Patients with Cirrhosis in the NIS 1998-2003 by Race and Ethnicity
Clinical VariableWhite (N = 41,633)African American (N = 6,962)Hispanic (N = 11,888)Asian (N = 1,211)Other (N = 2,002)
  • NOTE. “Other” indicates Native American or other race.

  • *

    P < 0.001, compared to white patients.

  • P < 0.05, compared to white patients.

Ascites21,134 (50.7)3,106 (44.5)*5,277 (44.5)*503 (41.5)*900 (44.6)
Hepatic encephalopathy23,696 (56.9)4,429 (63.7)*7,592 (63.8)*734 (60.5)1,175 (58.9)
Hepatorenal syndrome1,145 (2.8)197 (2.9)305 (2.6)36 (3.0)53 (2.7)
Variceal bleed3,910 (9.3)585 (8.4)1,090 (9.2)119 (10.0)216 (10.7)
Charlson Index (SE)4.2 (0.01)4.3 (0.04)*4.3 (0.04)*4.7 (0.09)*4.1 (0.06)

Rates of Portosystemic Shunt Procedures.

In comparison to whites, AAs and Hispanics were less likely to undergo a portosystemic shunt procedure for complications of portal hypertension (28.7 versus 12.2 and 17.1 procedures, respectively, per 1,000 hospitalizations). After multivariate adjustment for age, Charlson Index, sex, clinical presentation of portal hypertension, type of health insurance, calendar year, geographic region, and hospital characteristics (Fig. 1), the odds ratios of undergoing portosystemic shunt for AAs and Hispanics were 0.37 (0.27-0.51) and 0.69 (0.54-0.88), respectively, compared to whites. Patients were less likely to undergo a shunt procedure if they received Medicare (OR 0.70; 95% CI: 0.60-0.83) or Medicaid (OR 0.63; 95% CI: 0.53-0.76) or if they were uninsured (OR 0.40; 95% CI: 0.28-0.58) than if they had private insurance. There was no statistically significant interaction between race and type of health insurance. There was some modest geographic variation in the use of shunt procedures, with patients in the Midwest 2-fold more likely to receive a shunt than were those in the Northeast.

thumbnail image

Figure 1. Sociodemographic and clinical predictors for receiving a portosystemic shunt in the Nationwide Inpatient Sample between 1998 and 2003. Solid boxes indicate adjusted odds ratios with 95% confidence intervals of receiving a portosytemic shunt. Each odds ratio was simultaneously adjusted for all other variables. A solid circle indicates the reference category for polychotamous variables. An odds ratios greater than 1 favors a portosystemic shunt.

Download figure to PowerPoint

Variceal Hemostasis Among Patients with Variceal Bleeding.

The rate of variceal bleeders undergoing upper endoscopy and variceal treatment was similar among whites (71.8%), AAs (70.4%), Hispanics (71.0%), and Asians (75%). However, a higher percentage of AAs than whites underwent delayed endoscopy, defined as more than 24 hours after admission (23% versus 17%, P = 0.01). The proportion of Hispanics and Asian/Pacific Islanders with variceal bleeding who had delayed endoscopy was similar to that of whites (Fig. 2). After multivariate analysis and adjustment for potential confounders, the odds ratio of AAs having delayed endoscopy compared to whites was 1.6 (95% CI: 1.2-2.1). The likelihood of waiting more than 24 hours to have an endoscopy was also greater for patients receiving Medicare (OR 1.7; 95% CI: 1.2-2.2) and Medicaid (OR 1.7; 95% CI: 1.3-2.2) than for those who had private insurance.

thumbnail image

Figure 2. Proportion of patients admitteds with variceal bleeding who experienced a delayed upper endoscopy (>24 hours) stratified by race. AAs were more likely than non-Hispanic whites to wait more than 24 hours after admission to undergo upper endoscopy for variceal bleeding.

Download figure to PowerPoint

Rates of Liver Transplantation.

Compared to the rate of liver transplantation in white patients (61.3 LTs/1,000 hospitalizations), the rate was lower in both AA patients (23.1 LTs/1,000 hospitalizations, P < 0.0001) and Hispanic patients (29.9 LTs/1,000 hospitalizations, P = 0.04), but was not significantly lower in patients who were Asian/Pacific Islander (42.9 LTs/1,000 hospitalizations, P = 0.4). Multivariate analysis showed the likelihood of undergoing an LT was considerably lower for AAs (OR 0.32; 95% CI: 0.20-0.52) and Hispanics (OR 0.46; 95% CI: 0.25-0.83) than for whites after adjusting for confounders. A subgroup analysis of admissions prior to incorporation of the MELD score for liver allocation showed persistent findings that AAs (OR 0.42; 95% CI: 0.26-0.66) and Hispanics (OR 0.51; 95% CI: 0.27-0.96) were less likely to receive LTs than whites.

Type of health insurance was also a strong and independent predictor of undergoing liver transplantation (Fig. 3). Compared to those with private insurance, patients receiving Medicare (OR 0.42; 95% CI: 0.26-0.67) and Medicaid (OR 0.37; 95% CI: 0.27-0.50) were much less likely to receive a liver transplant. Likewise, being uninsured was associated with a lower rate of LT (OR 0.36; 95% CI: 0.18-0.72). There was no evidence of interaction between race and health insurance status. Hepatorenal syndrome was highly associated with likelihood of undergoing LT (OR 3.8; 95% CI: 2.2-6.5), whereas variceal bleeding was inversely associated with LT (OR 0.20; 95% CI: 0.07-0.53). There was no statistically significant geographic variation in LT.

thumbnail image

Figure 3. Sociodemographic and clinical predictors of receiving a liver transplant in the Nationwide Inpatient Sample between 1998 and 2003. Solid boxes indicate adjusted odds ratios with 95% confidence intervals of undergoing a liver transplantation. Each odds ratio was simultaneously adjusted for all other variables. Solid circles indicate the reference category for polychotamous variables. An odds ratios greater than 1 favors liver transplantation.

Download figure to PowerPoint

In-Hospital Mortality.

The overall in-hospital mortality rate was 8.4% of all hospitalizations for complications of portal hypertension. Relative to that of whites (8.4%), mortality of AAs was significantly higher (9.4%, P = 0.01) and of Hispanics significantly lower (7.2%, P < 0.001). These relative racial differences in mortality were preserved after multivariate adjustment for potential confounders. The adjusted odds ratio of death compared to whites was 1.12 for AAs (95% CI: 1.01-1.24) and 0.83 for Hispanics (95% CI: 0.75-0.92). Uninsured patients had higher mortality than those with private insurance (OR 1.64; 95% CI: 1.4-1.9). There was no interaction between the effects of race and health insurance on in-hospital mortality. Having hepatorenal syndrome was the strongest predictor of death (OR 9.4; 95% CI: 8.3-10.5). Adjusted mortality was lower in the West than in the Northeast (OR 0.80; 95% CI: 0.71-0.90). Racial differences in mortality were more pronounced among patients who had presented with variceal bleeding. In this subgroup, AAs were more likely than whites to die in the hospital (OR 1.72; 95% CI: 1.23-2.39).

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

We have shown striking racial variation in the United States in the use of surgical and endoscopic procedures in the inpatient management of complications of portal hypertension. Though the reasons for these racial differences are unclear from this study, the influence of health insurance on rates of portosystemic shunts and LT suggests that nonmedical factors such as health care access may play a contributing role. The presence of racial differences among NIS hospital discharges reflects potential nationwide health disparities in the treatment of a chronic condition that is among the leading causes of death in this nation.

We are the first to report racial variation in receiving a portosystemic shunt and endoscopic variceal hemostasis for complications of portal hypertension using a national database. Though the indications for portosystemic shunts may vary depending on the manifestation of portal hypertension, racial differences were persistent even after adjusting for disease presentation. Because we were not able to control for disease severity, these findings can only be interpreted cautiously as exploratory evidence of racial disparities. It is possible that AAs and Hispanics respond more readily to endoscopic hemostasis and do not require portosystemic shunt as frequently as whites, but this is only a hypothetical explanation and has not been observed in any previous intervention trials. The higher rate of portosystemic shunts among those with private insurance, however, does suggest socioeconomic factors such as access to physicians or high quality of hospital care may be contributors. Both radiological and surgical shunt procedures have been shown to be effective in reducing rebleeding from varices and in decreasing mortality and improving quality of life.19 Thus, the potential for disparities in palliative shunt procedures warrants further investigation.

Similarly, prompt endoscopic hemostasis is considered first-line therapy for variceal bleeding.20 Though published guidelines have not established formal recommendations for the timing of endoscopy, a recent survey showed that three-quarters of gastroenterologists believed emergent endoscopy should be performed within 12 hours, and all those surveyed thought upper endoscopy should be done in less than 24 hours.21 Our findings of more frequent delays in endoscopy being performed for AAs than for whites suggest several interpretations. Upper endoscopy is such a widely performed procedure that racial differences in access to a gastroenterologist is not expected to be a major issue, especially after adjustment for rural location. This supposition is also supported by there not having been racial differences in the proportion of variceal bleeders who underwent endoscopy at some point during an admission. Delayed endoscopy for AAs may reflect lower disease severity, though the higher mortality of AA variceal bleeders would argue against this interpretation. Because endoscopic hemostasis was delayed only in AAs and not in other minorities, culture-specific preferences not to undergo invasive procedures may be at work. Finally, these observations may reflect racially-based variations in physician threshold for performing upper endoscopy that betray disparities in quality of care.

Reid et al. previously demonstrated racial disparity in liver transplantation in the UNOS system, which comprises patients awaiting LT.7 Their study offered evidence that AAs were underrepresented on the transplantation list relative to the general population and were less likely to receive an LT within 4 years of being placed on the list. In reality, findings from the UNOS study population likely underestimate the extent of disparities because this selected population has already overcome significant barriers including medical and social evaluations that are required to be listed for LT. The NIS data allows a more comprehensive exploration of potential racial disparities in LT that includes individuals who have not been successfully listed. Except for a few with Childs A cirrhosis, most hospitalized patients with cirrhosis and complications of portal hypertension have a severity of disease considered an appropriate indication for LT. Unfortunately, because the NIS does not have a variable indicating whether a patient is listed for LT, we were unable to characterize the contribution of racial differences in being listed to overall disparities.

A variety of reasons could explain why some patients who had clinical indications for LT were not listed, including ongoing substance abuse, lack of social support, and comorbidities. Patients with appropriate indications for LT may also have been discharged to await LT as outpatients. One limitation of the NIS is that it did not allow us to follow these patients, and thus our analysis does not include LTs performed after discharge. However, for this alternative explanation to bias the strong associations observed, there would have to be large racial disparities in the manner that AAs and whites await transplant, with many AA patients discharged to await LT as outpatients, but most whites remaining in the hospital to await LT. This scenario is unlikely. It is more plausible that racial differences in LT reflect underlying social mechanisms that are also responsible for variation in the utilization of portosystemic shunts and endoscopic hemostasis. Shunts and endoscopy are done during the same admission in patients with acute variceal bleeding, and lack of longitudinal data should not bias observations of disparity for those interventions.

The potential mechanisms for racial disparities observed in our study are likely multifactorial. Data reported in the renal transplantation literature suggest that disease severity and comorbidity,22 low referral rate,23, 24 patient preference,24, 25 and socioeconomic factors26, 27 may all contribute to disparities in solid-organ transplantation. Even though hospitalization is a crude indicator of disease severity, one limitation of the NIS is that it does not contain data that allows direct assessment of severity using a liver-disease-specific scale such as the model for end-stage liver disease (MELD), a validated predictor of 3-month mortality.18 The Charlson Index is widely used to account for disease burden in administrative data. However, it has been validated for general in-hospital mortality and has not been modified specifically for liver disease. Thus, we cannot exclude the possibility that racial differences in in-hospital mortality may have resulted from differences in disease severity that were not adequately controlled for by the Charlson Index or variables for clinical disease presentation.

The MELD score, however, was unlikely to influence LT outcomes in this study population because it was not approved for donor liver allocation until February 27, 2002. More than two-thirds of the admissions in this study were prior to the MELD era, during which decisions on prioritization for LT were based on Child-Turcotte-Pugh (CTP) scores and waiting list times. Our sensitivity analysis that included only admissions in the pre-MELD era (before 2002) yielded similar findings. Thus, for the purposes of accounting for disease severity and appropriate indication with respect to LT, the use of portal hypertension complications (e.g., encephalopathy and ascites), 2 of which are components of the CTP score, was reasonable. Nonetheless, we must cautiously interpret these findings of disparities as preliminary, requiring confirmation through prospectively collected clinical data.

Our data suggest that health insurance (one indicator of health care access) is associated with likelihood of liver transplantation. However, the racial differences observed in this study were independent of type of health insurance. We could speculate that possible racial differences in sociodemographic factors are determinants of LT. However, racial differences in portosystemic shunt procedures and variceal hemostasis would not be as contingent on having a family and social support as they are for LT. Further studies to address these issues and the mechanisms of racial disparities are clearly warranted.

The lower mortality of Hispanics than whites despite less resource utilization may point to lesser disease severity in this ethnic group. However, the higher mortality of African Americans raises concern that disparities in inpatient management of decompensated cirrhosis are affecting overall outcomes. Our findings of racial differences in mortality from all causes of end-stage liver disease extends previously described reports of higher in-hospital mortality among AAs with hepatitis C.28 Particularly concerning is the higher proportion of AA variceal bleeders than white variceal bleeders who undergo upper endoscopy more than 24 hours after admission. Consequently, the mortality risk is 70% higher for AAs compared to that of whites in this subgroup. Thus, characterization and elimination of the intermediary barriers that hinder prompt endoscopy may represent an opportunity to reduce racial differences in mortality.

One potential criticism of the present study is that we were unable to link a subsample of the NIS with medical records in order to validate ICD-9 coding. However, the NIS databases have been extensively validated in the last several years in comparison reports and are a considered a valuable tool in clinical research. Furthermore, any misclassification from inaccuracies in diagnostic ICD-9-CM codes would be distributed uniformly across all races and other demographic factors and therefore are nondifferential with respect to these variables. Nondifferential misclassification results in dilution of the effects by race and predictor variables. However, given that in our study we found strong associations between race and surgical and endoscopic procedures for complications of portal hypertension, we do not believe nondifferential misclassification would dramatically compromise our conclusions.29

We have previously shown that even when AAs receive liver transplantation, their outcomes are worse than those of whites, which could not be explained by disparities in socioeconomic status, education, or insurance.30–32 Our observations of racial disparities in the inpatient management of decompensated cirrhosis have implications for both clinical practice and future studies. Our study and other similar studies suggest that AAs are less likely to be listed and even when they are listed, less likely to receive transplantation in a timely fashion.7 Given the rapidly expanding minority populations, which are expected to become the majority of the population by 2050, addressing health disparities in cirrhosis, which has a big effect on both resource utilization and mortality, will be critical in the next several years. The NIS is a large and nationally representative data set that has enabled us to provide descriptive insight into areas that should be investigated in greater depth. However, given the limitations of the NIS data sets, these findings should be considered exploratory. Future primary studies are warranted to confirm and to elucidate the mechanisms of racial disparities in order to enact interventions to rectify them. Concurrently, it is a measure of good practice and quality of care to develop more standardized protocols in the management of portal hypertension to ensure equitable care regardless of race, health insurance coverage, or socioeconomic status.

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References