[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.161.216.242. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Download PDF
Figure 1.
Adjusted mean values of serum creatinine (A), estimated glomerular filtration rate (B), serum albumin (C), and hematocrit (D) by racial/ethnic group and quartile of socioeconomic status (SES) score (low to high socioeconomic groups, represented from left to right within each racial/ethnic grouping). Values are based on multivariate analysis of variance models that included the covariates of age, sex, comorbid conditions, type of medical insurance, employment status, ability to ambulate, and ability to transfer and an interaction term between race/ethnicity and quartile of SES score. Error bars represent 95% confidence intervals. To convert creatinine to micromoles per liter, multiply by 88.4.

Adjusted mean values of serum creatinine (A), estimated glomerular filtration rate (B), serum albumin (C), and hematocrit (D) by racial/ethnic group and quartile of socioeconomic status (SES) score (low to high socioeconomic groups, represented from left to right within each racial/ethnic grouping). Values are based on multivariate analysis of variance models that included the covariates of age, sex, comorbid conditions, type of medical insurance, employment status, ability to ambulate, and ability to transfer and an interaction term between race/ethnicity and quartile of SES score. Error bars represent 95% confidence intervals. To convert creatinine to micromoles per liter, multiply by 88.4.

Figure 2.
Adjusted mean values of serum creatinine (A), estimated glomerular filtration rate (B), serum albumin (C), and hematocrit (D) by racial/ethnic group and type of medical insurance. Values are based on multivariate analysis-of-variance models that included the covariates of age, sex, quartile of socioeconomic status score, comorbid conditions, employment status, ability to ambulate, and ability to transfer and an interaction term between race/ethnicity and type of medical insurance. Error bars represent 95% confidence intervals. To convert creatinine to micromoles per liter, multiply by 88.4.

Adjusted mean values of serum creatinine (A), estimated glomerular filtration rate (B), serum albumin (C), and hematocrit (D) by racial/ethnic group and type of medical insurance. Values are based on multivariate analysis-of-variance models that included the covariates of age, sex, quartile of socioeconomic status score, comorbid conditions, employment status, ability to ambulate, and ability to transfer and an interaction term between race/ethnicity and type of medical insurance. Error bars represent 95% confidence intervals. To convert creatinine to micromoles per liter, multiply by 88.4.

Table 1. 
Values for Serum Creatinine, Estimated GFR, Serum Albumin, and Hematocrit Using Either Educational Level or SES Score as the Measure of SES in Dialysis Morbidity and Mortality Wave II Study Participants
Values for Serum Creatinine, Estimated GFR, Serum Albumin, and Hematocrit Using Either Educational Level or SES Score as the Measure of SES in Dialysis Morbidity and Mortality Wave II Study Participants
Table 2. 
Values for Serum Creatinine, Estimated GFR, Serum Albumin, and Hematocrit by Race/Ethnicity, Type of Medical Insurance, and SES Score in Patients With End-Stage Renal Disease Due to Diabetes Mellitus or Lupus Nephritis*
Values for Serum Creatinine, Estimated GFR, Serum Albumin, and Hematocrit by Race/Ethnicity, Type of Medical Insurance, and SES Score in Patients With End-Stage Renal Disease Due to Diabetes Mellitus or Lupus Nephritis*
Table 3. 
Association of Race/Ethnicity, SES, and Type of Medical Insurance With Epoietin Use Before Dialysis in All Patients and in the Subgroup With Hematocrit Readings Less Than 30%
Association of Race/Ethnicity, SES, and Type of Medical Insurance With Epoietin Use Before Dialysis in All Patients and in the Subgroup With Hematocrit Readings Less Than 30%
1.
United States Renal Data System, USRDS 2005 Annual Data Report: Atlas of End-stage Renal Disease in the United States.  Bethesda, Md National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases2006;
2.
Ratcliffe  PJPhillips  REOliver  DO Late referral for maintenance dialysis. Br Med J (Clin Res Ed) 1984;288441- 443
PubMedArticle
3.
Jungers  PZingraff  JAlbouze  G  et al.  Late referral to maintenance dialysis: detrimental consequences. Nephrol Dial Transplant 1993;81089- 1093
PubMed
4.
Roubicek  CBrunet  PHuiart  L  et al.  Timing of nephrology referral: influence on mortality and morbidity. Am J Kidney Dis 2000;3635- 41
PubMedArticle
5.
Roderick  PJones  CDrey  N  et al.  Late referral for end-stage renal disease: a region-wide survey in the south west of England. Nephrol Dial Transplant 2002;171252- 1259
PubMedArticle
6.
Kinchen  KSSadler  JFink  N  et al.  The timing of specialist evaluation in chronic kidney disease and mortality. Ann Intern Med 2002;137479- 486
PubMedArticle
7.
Kessler  MFrimat  LPanescu  VBriançon  S Impact of nephrology referral on early and midterm outcomes in ESRD: EPidémiologie de l’Insuffisance REnale chronique terminale en Lorranine (EPIREL): results of a 2-year, prospective, community-based study. Am J Kidney Dis 2003;42474- 485
PubMedArticle
8.
Stack  AG Impact of timing of nephrology referral and pre-ESRD care on mortality risk among new ESRD patients in the United States. Am J Kidney Dis 2003;41310- 318
PubMedArticle
9.
Obialo  CIOfili  EOQuarshie  AMartin  PC Ultralate referral and presentation for renal replacement therapy: socioeconomic implications. Am J Kidney Dis 2005;46881- 886
PubMedArticle
10.
Ifudu  ODawood  MHomel  PFriedman  EA Excess morbidity in patients starting uremia therapy without prior care by a nephrologist. Am J Kidney Dis 1996;28841- 845
PubMedArticle
11.
Sesso  RYoshihiro  MM Time of diagnosis of chronic renal failure and assessment of quality of life in hemodialysis patients. Nephrol Dial Transplant 1997;122111- 2116
PubMedArticle
12.
Arora  PObrador  GTRuthazer  R  et al.  Prevalence, predictors, and consequences of late nephrology referral at a tertiary care center. J Am Soc Nephrol 1999;101281- 1286
PubMed
13.
Obrador  GTRuthazer  RArora  PKausz  ATPereira  BJG Prevalence of and factors associated with suboptimal care before initiation of dialysis in the United States. J Am Soc Nephrol 1999;101793- 1800
PubMed
14.
Curtis  BMBarrett  BJJindal  K  et al.  Canadian survey of clinical status at dialysis initiation 1998-1999: a multicenter prospective survey. Clin Nephrol 2002;58282- 288
PubMed
15.
Lorenzo  VMartin  MRufino  MHernández  DTorres  AAyus  JC Predialysis nephrologic care and a functioning arteriovenous fistula at entry are associated with better survival in incident hemodialysis patients: an observational cohort study. Am J Kidney Dis 2004;43999- 1007
PubMedArticle
16.
Churchill  DNTaylor  DWCook  RJ  et al.  Canadian hemodialysis mortality study. Am J Kidney Dis 1992;19214- 234
PubMedArticle
17.
Collins  AJMa  JZUmen  AKeshaviah  P Urea index and other predictors of hemodialysis patient survival. Am J Kidney Dis 1994;23272- 282
PubMedArticle
18.
Foley  RNParfrey  PSHarnett  JDKent  GMMurray  DCBarre  PE Hypoalbuminemia, cardiac morbidity, and mortality in end-stage renal disease. J Am Soc Nephrol 1996;7728- 736
PubMed
19.
Fink  JCBurdick  RAKurth  SJ  et al.  Significance of serum creatinine values in end-stage renal disease patients. Am J Kidney Dis 1999;34694- 701
PubMedArticle
20.
Li  SCollins  AJ Association of hematocrit value with cardiac morbidity and mortality in incident hemodialysis patients. Kidney Int 2004;65626- 633
PubMedArticle
21.
Cooper  BAPenne  ELBartlett  LHPollock  CA Protein malnutrition and hypoalbuminemia as predictors of vascular events and mortality in ESRD. Am J Kidney Dis 2004;4361- 66
PubMedArticle
22.
Regidor  DLKopple  JDKovesdy  CP  et al.  Associations between changes in hemoglobin and administered erythopoiesis-stimulating agent and survival in hemodialysis patients. J Am Soc Nephrol 2006;171181- 1191
PubMedArticle
23.
Frankenfield  DLRocco  MVFrederick  PRPugh  JMcClellan  WMOwen  WF  JrNational ESRD Core Indicators Work Group, Racial/ethnic analysis of selected intermediate outcomes for hemodialysis patients: results from the 1997 ESRD Core Indicators Project. Am J Kidney Dis 1999;34721- 730
PubMedArticle
24.
Sehgal  AR Impact of quality improvement efforts on race and sex disparities in hemodialysis. JAMA 2003;289996- 1000
PubMedArticle
25.
Alexander  GCSehgal  AR Barriers to cadaveric renal transplantation among blacks, women, and the poor. JAMA 1998;2801148- 1152
PubMedArticle
26.
Epstein  AMAyanian  JZKeogh  JH  et al.  Racial disparities in access to renal transplantation: clinically appropriate or due to underuse or overuse? N Engl J Med 2000;3431537- 1544
PubMedArticle
27.
Wolfe  RAAshby  VBMilford  EL  et al.  Differences in access to cadaveric renal transplantation in the United States. Am J Kidney Dis 2000;361025- 1033
PubMedArticle
28.
Yeates  KESchaubel  DECass  ASequist  TDAyanian  JZ Access to renal transplantation for minority patients with ESRD in Canada. Am J Kidney Dis 2004;441083- 1089
PubMedArticle
29.
Ifudu  ODawood  MIofel  YValcourt  JSFriedman  EA Delayed referral of black, Hispanic, and older patients with chronic renal failure. Am J Kidney Dis 1999;33728- 733
PubMedArticle
30.
Iofel  YDawood  MValcourt  JSIfudu  O Initiation of dialysis is not delayed in whites with progressive renal failure. ASAIO J 1998;44M598- M600
PubMedArticle
31.
Obrador  GTArora  PKraus  ATRuthazer  RGereira  BJGLevey  AS Level of renal function at the initiation of dialysis in the U.S. end-stage renal disease population. Kidney Int 1999;562227- 2235
PubMedArticle
32.
Kausz  ATObrador  GTArora  PRuthazer  RLevey  ASPereira  BJG Late initiation of dialysis among women and ethnic minorities in the United States. J Am Soc Nephrol 2000;112351- 2357
PubMed
33.
Soucie  JMNeylan  JFMcClellan  W Race and sex differences in the identification of candidates for renal transplantation. Am J Kidney Dis 1992;19414- 419
PubMedArticle
34.
Kasiske  BLLondon  WEllison  MD Race and socioeconomic factors influencing early placement on the kidney transplant waiting list. J Am Soc Nephrol 1998;92142- 2147
PubMed
35.
Levey  ASGreene  TKusek  JWBeck  GJ A simplified equation to predict glomerular filtration rate from serum creatinine [abstract]. J Am Soc Nephrol 2000;11A0828
36.
Yost  KPerkins  CCohen  RMorris  CWright  W Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control 2001;12703- 711
PubMedArticle
37.
Diez-Roux  AVKiefe  CIJacobs  DR  Jr  et al.  Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies. [published correction appears in Ann Epidemiol. 2001;30:924] Ann Epidemiol 2001;11395- 405
PubMedArticle
38.
 Census 2000 Summary File 3—United States. http://factfinder.census.gov. Accessed December 22, 2005
39.
Byrne  CNedelman  JLuke  RG Race, socioeconomic status, and the development of end-stage renal disease. Am J Kidney Dis 1994;2316- 22
PubMedArticle
40.
Perneger  TVWhelton  PKKlag  MJ Race and end-stage renal disease: socioeconomic status and access to health care as mediating factors. Arch Intern Med 1995;1551201- 1208
PubMedArticle
41.
Klag  MJWhelton  PKRandall  BLMeaton  JDBrancati  FLStamler  J End-stage renal disease in African-American and White men: 16-year MRFIT findings. JAMA 1997;2771293- 1298
PubMedArticle
42.
Kirk  RE Experimental Design: Procedures for the Behavioral Sciences. 2nd ed. Belmont, Calif Brooks/Cole1982;
43.
Doescher  MPSaver  BGFranks  PFiscella  K Racial and ethnic disparities in perceptions of physician style and trust. Arch Fam Med 2000;91156- 1163
PubMedArticle
44.
Berrios-Rivera  JPStreet  RL  JrGarcia Popa-Lisseanu  MG  et al.  Trust in physicians and elements of the medical interaction in patients with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum 2006;55385- 393
PubMedArticle
45.
Owen  WF  JrSzczech  LAFrankenfield  DL Healthcare system interventions for inequality in quality: corrective action through evidence-based medicine. J Natl Med Assoc 2002;94 ((suppl)) 83S- 91S
PubMed
46.
Boulware  LETroll  MUJaar  BGMyers  DIPowe  NR Identification and referral of patients with progressive CKD: a national study. Am J Kidney Dis 2006;48192- 204
PubMedArticle
47.
Fox  CHBrooks  AZayas  LEMcClellan  WMurray  B Primary care physicians' knowledge and practice patterns in the treatment of chronic kidney disease: an Upstate New York Practice-based Research Network (UNYNET) study. J Am Board Fam Med 2006;1954- 61
PubMedArticle
48.
Abuelo  JGShemin  DChazan  JA Serum creatinine concentration at the onset of uremia: higher levels in black males. Clin Nephrol 1992;37303- 307
PubMed
49.
Cody  JDaly  CCampbell  M  et al.  Recombinant human erythropoietin for chronic renal failure anaemia in pre-dialysis patients. Cochrane Database Syst Rev 2005;3CD003266
PubMed
50.
Obrador  GTRoberts  TSt. Peter  WLFrazier  EPereira  BJCollins  AJ Trends in anemia at initiation of dialysis in the United States. Kidney Int 2001;601875- 1884
PubMedArticle
51.
Krieger  NChen  JTWaterman  PDSoobader  M-JSubramanian  SVCarson  R Choosing area based socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning: the Public Health Disparities Geocoding Project (US). J Epidemiol Community Health 2003;57186- 199
PubMedArticle
Original Investigation
May 28, 2007

Laboratory Abnormalities at the Onset of Treatment of End-Stage Renal DiseaseAre There Racial or Socioeconomic Disparities in Care?

Author Affiliations

Author Affiliation: Intramural Research Program, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Md (Dr Ward).

Arch Intern Med. 2007;167(10):1083-1091. doi:10.1001/archinte.167.10.1083
Abstract

Background  Laboratory abnormalities at the start of treatment of end-stage renal disease (ESRD) have been reported as worse in racial/ethnic minorities than in white patients, suggesting racial disparities in care. It is not known whether these differences are attributable to racial/ethnic differences in socioeconomic status (SES).

Methods  We tested associations between race/ethnicity, SES, and type of medical insurance and serum creatinine level, estimated glomerular filtration rate, serum albumin level, and hematocrit at the start of treatment of ESRD and use of epoietin before ESRD treatment in a large national population-based sample. Data on 515 561 patients beginning ESRD treatment between January 1, 1996, and June 30, 2004, were obtained for this cross-sectional survey from the United States Renal Data System.

Results  Race/ethnicity had a much stronger association than SES with each laboratory measure. Adjusted mean serum creatinine levels were lowest in white patients (7.5 mg/dL [663.0 μmol/L]; 95% confidence interval [CI], 7.45-7.49) and highest in black patients (8.9 mg/dL [786.7 μmol/L]; 95% CI, 8.92-8.97) (P<.001 across racial/ethnic groups). Adjusted mean hematocrit for white patients (29.5%; 95% CI, 29.4%-29.6%) was significantly higher and for black patients (28.3%; 95% CI, 28.2%-28.4%) significantly lower than that of all other racial/ethnic groups (P<.001 across racial/ethnic groups). Less marked differences were present for estimated glomerular filtration rate and serum albumin level. In contrast, predialysis use of epoietin was associated with race/ethnicity (black vs white: odds ratio, 0.80; 95% CI, 0.78-0.81; Hispanic vs white: odds ratio, 0.87; 95% CI, 0.85-0.89) and showed a graded decrease with decreasing SES(odds ratio for the lowest vs highest socioeconomic quartile 0.68; 95% CI, 0.67-0.70). Patients without medical insurance had more abnormal laboratory values than those with insurance, but these associations were weaker than those of race/ethnicity.

Conclusions  Minorities, particularly black patients, had more severe laboratory abnormalities at the start of ESRD treatment than white patients. These differences were not readily attributable to SES differences. Absence of medical insurance, SES, and race/ethnicity were associated with the likelihood of predialysis use of epoietin.

More than 100 000 patients begin treatment for end-stage renal disease (ESRD) in the United States annually.1 The timing of initiation of renal replacement treatment has a major impact on the future health of patients. Those who begin treatment on an urgent or emergency basis have higher risks of complications, hospitalization, and mortality than those whose treatment is planned.29 Patients receiving emergency treatment also have higher serum creatinine levels, worse anemia, and more severe malnutrition at the start of renal replacement treatment.315 The degree of abnormality in laboratory test values at the start of renal replacement treatment may, thus, serve as a measure of the quality of care of patients with chronic kidney disease. Anemia and hypoalbuminemia at the start of dialysis have also been associated with increased morbidity and mortality risks.4,5,8,9,1622

Racial/ethnic disparities have been reported for the dose of hemodialysis and for access to renal transplantation.2328 Similar disparities may exist for predialysis care. Late referral to nephrologic care is more common among racial/ethnic minorities,6,29 and minorities have higher serum creatinine levels, worse anemia, and lower serum albumin levels at the start of renal replacement treatment than white patients.13,19,2932 However, it is not clear whether these racial/ethnic differences reflect only differences in socioeconomic status (SES). Several studies12,13,19,31,32 used medical insurance or employment status as measures of SES, which may not adequately measure SES.

The objective of this study is to determine whether any race/ethnicity-associa ted differences in laboratory measures at the start of dialysis can be accounted for by differences in SES or medical insurance status. Predialysis use of epoietin is also examined as a measure of disparities in the treatment of patients with chronic kidney disease.

METHODS
DATA AND PATIENTS

Information on patients with incident ESRD was obtained from the United States Renal Data System (USRDS), a national population-based registry of all patients with ESRD.1 Patients are enrolled in the USRDS after being certified as needing long-term renal replacement therapy by their nephrologist. The USRDS includes patient demographic and clinical characteristics and selected laboratory test values at the start of renal replacement therapy, types of renal replacement therapy, and outcomes. The study protocol was exempted from human subjects review by the National Institutes of Health Office of Human Subjects Research.

Data were abstracted on all patients with incident ESRD between January 1, 1996, and June 30, 2004, who resided in 1 of the 50 states or the District of Columbia (N = 779 918). We excluded 2001 patients with missing data on age, sex, or race and 20 641 (2.65%) with missing or invalid ZIP codes, leaving 757 276 patients. We then limited the analysis to the 747 556 patients who were 20 years or older to minimize confounding by age-associated differences in laboratory values and patterns of care. We also excluded 12 854 patients who had undergone renal transplantation within 30 days of beginning ESRD treatment because these patients are generally healthier and wealthier than those treated with dialysis.25,26,33,34 Last, we excluded 219 141 patients who had missing values for serum creatinine or whose reported values were obtained more than 45 days before or more than 1 day after the start of dialysis. The final sample consisted of 515 561 patients.

STUDY VARIABLES

Four laboratory values were studied as dependent variables: serum creatinine, serum albumin, hematocrit, and estimated glomerular filtration rate (GFR). The GFR was calculated using the 4-variable model of the Modification of Diet in Renal Disease study.35 We also examined predialysis use of epoietin.

The independent variables of interest were race/ethnicity, SES, and medical insurance status. Medical insurance before renal replacement therapy was categorized as none, Medicare only, Medicaid with or without Medicare, and private insurance with or without Medicare. The USRDS does not include patient-level measures of SES. Therefore, we developed a composite measure of SES and assigned an SES score to each patient based on the characteristics of their ZIP code of residence.36,37 We first abstracted data from the 2000 US Census on income, wealth, education, and occupation according to ZIP code tabulation area.38 By using principal components analysis we identified 7 census measures to include in the SES measure. Each of these measures loaded strongly on a single factor, with all factor loads being greater than 0.75, and together they explained 70% of the variance across ZIP codes. We then computed z scores for each ZIP code for each measure. The composite SES score was then computed as the sum of the z scores for all 7 measures.

To validate this measure, we compared SES scores across educational levels in 2394 patients who participated in the USRDS Dialysis Morbidity and Mortality Wave II substudy,1 which collected information on educational attainment. Mean SES scores were 0, 0.80, 1.35, and 2.85 for those with fewer than 12 years of school, high school graduates, some college, and college graduates, respectively (P<.001 for linear trend).

Covariates included patient age, sex, each of 11 comorbid conditions, employment status, and functional status. Employment was coded as present if the patient was employed full-time or part-time at the start of renal replacement therapy. Inability to ambulate and inability to transfer were used to measure functional status.

STATISTICAL ANALYSIS

Values of serum creatinine and GFR were log-transformed for analysis and then transformed back to natural units for presentation. The SES scores were categorized into quartiles to allow for nonlinear associations with the laboratory values and epoietin use. Because ESRD occurs more commonly among those of lower SES, the SES categories do not define quartiles of patients.3941

Univariable comparisons of laboratory values among racial/ethnic groups, SES score quartiles, and patients with different types of insurance coverage were performed using analysis of variance. Comparisons of predialysis epoietin use were performed using logistic regression analysis. Multivariate models included the covariates of patient age, sex, comorbid conditions, employment, and functional status. The interactions between race/ethnicity and SES score quartile and between race/ethnicity and insurance type were also tested. The magnitude of association between race/ethnicity, SES score, and type of medical insurance and each laboratory measure was assessed using ω2, a measure of effect size that represents the proportion of the dependent variance accounted for by an independent variable in the population.42 Although ω2 has a theoretical range of 0 to 1, values greater than 0.1 are uncommon, and values of 0.05 are considered to represent moderate effects.42

Analyses were first performed in the subgroup of patients who participated in the USRDS Dialysis Morbidity and Mortality Wave II study using either educational level or the composite SES score as the measure of SES. The main analysis used the SES score and included all patients with available laboratory data. Because considerations for initiating renal replacement therapy differ among primary renal diseases, we also tested associations in patients with ESRD due to diabetes mellitus, for whom treatment often begins at higher levels of renal function than for those with other causes of ESRD31 and those with ESRD due to lupus nephritis, a large proportion of whom are racial/ethnic minorities. All the analyses were performed using SAS programs (SAS Institute Inc, Cary, NC). All hypothesis testing was 2-tailed. Because of the large sample, most P values were small, and effect sizes (ω2) were considered more relevant for comparison.

RESULTS
ASSOCIATIONS IN THE DIALYSIS MORBIDITY AND MORTALITY WAVE II STUDY PARTICIPANTS

Of the 2394 patients (mean ± SD age, 57.5 ± 15.3 years; 53.05% men) who had values for serum creatinine and estimated GFR at the start of renal replacement therapy, 1883 (78.65%) also had values for serum albumin and 2169 (90.60%) also had hematocrit recorded in the same time frame. There were significant differences among racial/ethnic groups in serum creatinine values but no differences by level of education (Table 1). In the multivariate-adjusted model, white patients had significantly lower mean serum creatinine values than black patients(P<.001), Hispanic patients (P<.001), or American Indians (P = .04). Mean GFRs differed among racial/ethnic groups but not by educational level. Serum albumin levels did not vary significantly by race/ethnicity or educational level. Hematocrit readings were significantly higher in white patients than in black patients and Hispanic patients (P<.001 for both) but did not differ by educational level. Mean serum creatinine values were higher and mean GFR and hematocrit readings were lower in patients without medical insurance than in those with insurance (P<.05 for all) but did not differ among those with different types of insurance. There were no interactions between race/ethnicity and educational level or type of medical insurance (P>.20 for all). Comparison of ω2 values confirmed that race/ethnicity was more strongly associated with variations in each laboratory test than with educational level or medical insurance.

Adjustment using the composite SES score resulted in mean values for each racial/ethnic group that were similar to those based on adjustment using educational level as the measure of SES, supporting the validity of the composite SES score.

ASSOCIATIONS IN THE MAIN SAMPLE

The main sample (n = 515 561) included 271 390 white patients, 158 886 black patients, 58 701 Hispanic patients, 18 394 Asian/Pacific Islanders, 5587 American Indians, and 2603 persons of other race/ethnicity. Their mean ± SD age was 61.3 ± 15.3 years, and 54.03% were men. Of the sample, 69.42% had serum albumin values recorded in the 45 days before dialysis (n = 357 933), and 88.45% had hematocrit readings recorded (n = 456 010). Compared with patients included in the study, those who were excluded because serum creatinine values were missing or were outside the required time frame were older (mean ± SD age, 67.6 ± 14.2 years), had higher mean SES scores (0.9 vs 0.8), and included slightly fewer men (52.22%) and more white patients (69.82%).

Variations in serum creatinine levels were much greater for race/ethnicity (ω2 = 0.021) than for SES (ω2 = 0.0001) (Figure 1A) and type of medical insurance (ω2 = 0.006) (Figure 2A). Multivariate-adjusted mean serum creatinine levels were lowest in white patients (7.5 mg/dL [663.0 μmol/L]) and highest in black patients (8.9 mg/dL [786.7 μmol/L]). Mean serum creatinine levels were highest in those without medical insurance and those with missing data on insurance in each racial/ethnic group (Figure 2A).

Estimated GFRs at the onset of dialysis varied weakly with race/ethnicity (ω2 = 0.003) (Figure 1B and 2B) and type of medical insurance (ω2 = 0.0026) (Figure 2B) but were not associated with SES (Figure 1B). Mulitvariate-adjusted rates were highest in white patients (8.9 mL/min per 1.73 m2) and black patients (8.8 mL/min per 1.73 m2) and lowest in Asian/Pacific Islanders (7.6 mL/min per 1.73 m2). Estimated GFRs were lowest in those without medical insurance and in those with missing data on insurance in each racial/ethnic group.

Serum albumin levels varied significantly with race/ethnicity, SES (Figure 1C), and type of medical insurance (Figure 2C), but all associations were weak. White patients had the highest multivariate-adjusted levels (3.2 g/dL), and American Indians had the lowest levels (3.0 g/dL). Levels increased slightly with higher SES.

Hematocrit readings at the start of dialysis were also more strongly associated with race/ethnicity (ω2 = 0.006) than with SES (ω2 = 0.00004) (Figure 1D) or type of medical insurance (ω2 = 0.0029) (Figure 2D). The multivariate-adjusted mean hematocrit of white patients (29.5%) was significantly higher than that of all other racial/ethnic groups, and the value for black patients (28.3%) was significantly lower (P<.001 for all). Adjusted mean hematocrit increased slightly with increasing SES scores, from 28.6% in the lowest SES group to 29.0% in the highest SES group. Mean hematocrit readings were lowest in patients without medical insurance and those with missing data on insurance in each racial/ethnic group.

White-black differences in serum creatinine values and estimated GFRs at the start of dialysis were fairly uniform across ESRD networks. There was greater regional heterogeneity in white-Hispanic differences, which ranged from an adjusted mean difference in serum creatinine values of −0.1 mg/dL (−8.8 μmol/L) in network 9 (Indiana, Kentucky, and Ohio) to −1.24 mg/dL (−109.6 μmol/L) in network 6 (North Carolina, South Carolina, and Georgia) and an adjusted mean difference in estimated GFRs of 0.4 mL/min per 1.73 m2 in network 18 (southern California) to 1.37 mL/min per 1.73 m2 in network 10 (Illinois). There was also regional heterogeneity in hematocrit readings at the start of dialysis. White-black differences in hematocrit ranged from 0.7% in network 18 (southern California) to 1.5% in network 2 (New York), and white-Hispanic differences ranged from 0.1% in network 9 (Indiana, Kentucky, and Ohio) to 1.3% in network 13 (Arkansas, Louisiana, and Oklahoma).

Racial/ethnic differences in serum creatinine levels, estimated GFRs, and serum albumin levels remained stable across time. For example, adjusted mean white-black differences in serum creatinine levels were −1.6 mg/dL (−141.4 μmol/L) in 1996-1998 and −1.4 mg/dL (−123.7 μmol/L) in 2002-2004. White-black differences in hematocrit decreased slightly from 1.3% in 1996-1998 to 0.9% in 2002-2004, but white-Hispanic differences did not change.

SUBGROUPS WITH ESRD DUE TO DIABETES MELLITUS OR LUPUS NEPHRITIS

The timing of ESRD treatment and the racial/ethnic composition of patients may differ for primary renal diseases, which could confound associations among race/ethnicity, SES, and laboratory values. To control for this possibility we examined associations among patients with ESRD due to diabetes mellitus or lupus nephritis. Associations in patients with ESRD due to diabetes mellitus (n = 208 266; mean ± SD age, 63.1 ± 11.7 years; 49.85% men) were similar to those in the main analysis (Table 2). White patients had higher mean serum creatinine values, and black patients had lower values, than those in other racial/ethnic groups. Estimated GFRs were highest in white patients and lowest in American Indians. Hematocrit readings varied more strongly by race/ethnicity than SES. Type of medical insurance was weakly associated with each laboratory measure.

Among patients with ESRD due to lupus nephritis (n = 6018; mean ± SD age, 41.8 ± 14.2 years; 82.49% women), results were again similar to those in the main analysis (Table 2). In this smaller sample, there was no association between SES and any laboratory value or between race/ethnicity and estimated GFR. Serum creatinine values were lowest in white patients and highest in black patients. Black patients had lower hematocrit readings than the other groups. Mean serum creatinine values were higher and mean GFRs and hematocrit readings were lower in patients without medical insurance.

USE OF EPOIETIN

Racial/ethnic and SES differences in hematocrit readings may be related to variations in the use of epoietin in the period before dialysis. There was a graded decrease in the likelihood of epoietin use with decreasing SES score, even with adjustment for medical insurance status (Table 3). Compared with white patients, epoietin use was slightly more likely among Asian/Pacific Islanders and less likely among other racial/ethnic groups. Similar findings were present in patients with a predialysis hematocrit less than 30%.

COMMENT

In this large study, mean serum creatinine levels at the start of dialysis were approximately 1.5 mg/dL (132.6 μmol/L) lower in white patients than in black patients, 1.0 mg/dL (88.4 μmol/L) lower in white patients than in Asian/Pacific Islanders, and 0.5 mg/dL (44.2 μmol/L) lower in white patients than in Hispanic patients and American Indians. Mean hematocrit readings were greater than 1% higher in white patients than in black patients and slightly less than 1% higher in white patients than in Hispanic patients, Asian/Pacific Islanders, and American Indians. These differences were present despite adjustment for SES and type of medical insurance. Measures of SES were only weakly associated with laboratory values. The effect sizes for race/ethnicity, although small, were much larger than those for SES and type of medical insurance, indicating the predominance of race/ethnicity in this relationship. These findings extend previous research12,13,19,31,32 that reported racial/ethnic differences in laboratory abnormalities at the start of dialysis but that did not include measures of SES or that examined insurance status only.

Racial/ethnic differences in laboratory values may have several different causes. The spectrum of primary renal diseases may differ among racial/ethnic groups. Diseases that are either asymptomatic, and, therefore, undetected, or rapidly progressive may occur more commonly in racial/ethnic minorities. However, results were similar among patients with ESRD due to diabetes mellitus or lupus nephritis, suggesting that differences in the distribution of primary diseases among racial/ethnic groups is not responsible for the racial/ethnic differences observed. Once chronic kidney disease is diagnosed, cultural differences in trust in physicians or the health care system may lead to delays in treatment.43,44 Differences in quality of care may also have a role.6,29 Deficiencies in patient-physician communication, patient education, and follow-up may contribute to racial/ethnic differences in laboratory abnormalities at the start of ESRD treatment.4547 Uremic symptoms are a major indication for dialysis, and black patients may be less likely than white patients to develop uremic symptoms at a given level of residual renal function.48 However, anemia and hypoalbuminemia belie the severity of chronic renal failure at the start of dialysis in black patients.

More specific evidence of racial/ethnic disparities in quality of care was the difference in predialysis use of epoietin. Black patients were 20% less likely and Hispanic patients and American Indians were 11% to 13% less likely than white patients to be treated with epoietin after adjustment for SES and medical insurance. Differences in the use of epoietin may contribute to differences in complications, exercise tolerance, and quality of life.49,50 Lower SES and the absence of medical insurance were also associated with lower likelihoods of predialysis use of epoietin, consistent with expectations for measures of access to quality care.

White-black differences in serum creatinine levels and estimated GFRs showed little regional variation, but there was substantial regional variation in hematocrit readings. There were also wide regional variations in white-Hispanic differences. This heterogeneity suggests that the factors responsible for racial/ethnic differences in laboratory abnormalities are not intrinsic cultural factors or related to minorities more likely declining treatment but rather that the factors are extrinsic and related to variations in quality of care. Studies that contrast processes of care across networks would be helpful in identifying ways to reduce these disparities. White-black differences in mean hematocrit readings decreased from 1996-1998 to 2002-2004, but few other changes indicated a decrease in racial/ethnic disparities across time.

The strengths of this study include the large population-based sample, testing for regional and temporal variations, and replication of results in 2 selected diseases. The study is limited in that personal measures of SES were not available for all patients. The main analysis used an area-based measure, which may misclassify patients.51 However, results were similar in the subgroup of patients for whom data on educational attainment were available. Data on laboratory values at the start of ESRD treatment were missing or outdated for some eligible patients, who were more likely to be older and white. This could bias the results if the missing values systematically differed from the values that were available, but this seems unlikely.

The results of this study indicate that the differences that exist among racial/ethnic groups in laboratory abnormalities at the start of ESRD treatment are likely not due to confounding by SES. Black patients had more abnormal values than patients of other racial/ethnic groups on several measures. Predialysis use of epoietin was lower among black patients, Hispanic patients, and American Indians; those of lower SES; and those without medical insurance. These differences, when considered in the context of previous evidence of late referral to nephrologic care, and historical differences in hemodialysis dose and in access to renal transplantation suggest racial/ethnic disparities in the quality of care of patients beginning renal replacement therapy. Understanding differences in the processes of care during this clinical transition may provide explanations for the origins of the disparity and may offer solutions.

Back to top
Article Information

Correspondence: Michael Ward, MD, MPH, Intramural Research Program, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, 10 Center Dr, Bldg 10 CRC, Room 4-1339, MSC 1468, Bethesda, MD 20892 (wardm1@mail.nih.gov).

Accepted for Publication: February 5, 2007.

Author Contributions: Dr Ward had full access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None reported.

Funding/Support: This research was supported by the Intramural Research Program of the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health.

Role of the Sponsor: The funding agency had no role in the design and conduct of this study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.

Disclaimer: The interpretation and reporting of these data are the responsibility of the author and in no way should be seen as an official policy or interpretation of the US federal government.

Acknowledgment: The data reported herein were supplied by the USRDS.

References
1.
United States Renal Data System, USRDS 2005 Annual Data Report: Atlas of End-stage Renal Disease in the United States.  Bethesda, Md National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases2006;
2.
Ratcliffe  PJPhillips  REOliver  DO Late referral for maintenance dialysis. Br Med J (Clin Res Ed) 1984;288441- 443
PubMedArticle
3.
Jungers  PZingraff  JAlbouze  G  et al.  Late referral to maintenance dialysis: detrimental consequences. Nephrol Dial Transplant 1993;81089- 1093
PubMed
4.
Roubicek  CBrunet  PHuiart  L  et al.  Timing of nephrology referral: influence on mortality and morbidity. Am J Kidney Dis 2000;3635- 41
PubMedArticle
5.
Roderick  PJones  CDrey  N  et al.  Late referral for end-stage renal disease: a region-wide survey in the south west of England. Nephrol Dial Transplant 2002;171252- 1259
PubMedArticle
6.
Kinchen  KSSadler  JFink  N  et al.  The timing of specialist evaluation in chronic kidney disease and mortality. Ann Intern Med 2002;137479- 486
PubMedArticle
7.
Kessler  MFrimat  LPanescu  VBriançon  S Impact of nephrology referral on early and midterm outcomes in ESRD: EPidémiologie de l’Insuffisance REnale chronique terminale en Lorranine (EPIREL): results of a 2-year, prospective, community-based study. Am J Kidney Dis 2003;42474- 485
PubMedArticle
8.
Stack  AG Impact of timing of nephrology referral and pre-ESRD care on mortality risk among new ESRD patients in the United States. Am J Kidney Dis 2003;41310- 318
PubMedArticle
9.
Obialo  CIOfili  EOQuarshie  AMartin  PC Ultralate referral and presentation for renal replacement therapy: socioeconomic implications. Am J Kidney Dis 2005;46881- 886
PubMedArticle
10.
Ifudu  ODawood  MHomel  PFriedman  EA Excess morbidity in patients starting uremia therapy without prior care by a nephrologist. Am J Kidney Dis 1996;28841- 845
PubMedArticle
11.
Sesso  RYoshihiro  MM Time of diagnosis of chronic renal failure and assessment of quality of life in hemodialysis patients. Nephrol Dial Transplant 1997;122111- 2116
PubMedArticle
12.
Arora  PObrador  GTRuthazer  R  et al.  Prevalence, predictors, and consequences of late nephrology referral at a tertiary care center. J Am Soc Nephrol 1999;101281- 1286
PubMed
13.
Obrador  GTRuthazer  RArora  PKausz  ATPereira  BJG Prevalence of and factors associated with suboptimal care before initiation of dialysis in the United States. J Am Soc Nephrol 1999;101793- 1800
PubMed
14.
Curtis  BMBarrett  BJJindal  K  et al.  Canadian survey of clinical status at dialysis initiation 1998-1999: a multicenter prospective survey. Clin Nephrol 2002;58282- 288
PubMed
15.
Lorenzo  VMartin  MRufino  MHernández  DTorres  AAyus  JC Predialysis nephrologic care and a functioning arteriovenous fistula at entry are associated with better survival in incident hemodialysis patients: an observational cohort study. Am J Kidney Dis 2004;43999- 1007
PubMedArticle
16.
Churchill  DNTaylor  DWCook  RJ  et al.  Canadian hemodialysis mortality study. Am J Kidney Dis 1992;19214- 234
PubMedArticle
17.
Collins  AJMa  JZUmen  AKeshaviah  P Urea index and other predictors of hemodialysis patient survival. Am J Kidney Dis 1994;23272- 282
PubMedArticle
18.
Foley  RNParfrey  PSHarnett  JDKent  GMMurray  DCBarre  PE Hypoalbuminemia, cardiac morbidity, and mortality in end-stage renal disease. J Am Soc Nephrol 1996;7728- 736
PubMed
19.
Fink  JCBurdick  RAKurth  SJ  et al.  Significance of serum creatinine values in end-stage renal disease patients. Am J Kidney Dis 1999;34694- 701
PubMedArticle
20.
Li  SCollins  AJ Association of hematocrit value with cardiac morbidity and mortality in incident hemodialysis patients. Kidney Int 2004;65626- 633
PubMedArticle
21.
Cooper  BAPenne  ELBartlett  LHPollock  CA Protein malnutrition and hypoalbuminemia as predictors of vascular events and mortality in ESRD. Am J Kidney Dis 2004;4361- 66
PubMedArticle
22.
Regidor  DLKopple  JDKovesdy  CP  et al.  Associations between changes in hemoglobin and administered erythopoiesis-stimulating agent and survival in hemodialysis patients. J Am Soc Nephrol 2006;171181- 1191
PubMedArticle
23.
Frankenfield  DLRocco  MVFrederick  PRPugh  JMcClellan  WMOwen  WF  JrNational ESRD Core Indicators Work Group, Racial/ethnic analysis of selected intermediate outcomes for hemodialysis patients: results from the 1997 ESRD Core Indicators Project. Am J Kidney Dis 1999;34721- 730
PubMedArticle
24.
Sehgal  AR Impact of quality improvement efforts on race and sex disparities in hemodialysis. JAMA 2003;289996- 1000
PubMedArticle
25.
Alexander  GCSehgal  AR Barriers to cadaveric renal transplantation among blacks, women, and the poor. JAMA 1998;2801148- 1152
PubMedArticle
26.
Epstein  AMAyanian  JZKeogh  JH  et al.  Racial disparities in access to renal transplantation: clinically appropriate or due to underuse or overuse? N Engl J Med 2000;3431537- 1544
PubMedArticle
27.
Wolfe  RAAshby  VBMilford  EL  et al.  Differences in access to cadaveric renal transplantation in the United States. Am J Kidney Dis 2000;361025- 1033
PubMedArticle
28.
Yeates  KESchaubel  DECass  ASequist  TDAyanian  JZ Access to renal transplantation for minority patients with ESRD in Canada. Am J Kidney Dis 2004;441083- 1089
PubMedArticle
29.
Ifudu  ODawood  MIofel  YValcourt  JSFriedman  EA Delayed referral of black, Hispanic, and older patients with chronic renal failure. Am J Kidney Dis 1999;33728- 733
PubMedArticle
30.
Iofel  YDawood  MValcourt  JSIfudu  O Initiation of dialysis is not delayed in whites with progressive renal failure. ASAIO J 1998;44M598- M600
PubMedArticle
31.
Obrador  GTArora  PKraus  ATRuthazer  RGereira  BJGLevey  AS Level of renal function at the initiation of dialysis in the U.S. end-stage renal disease population. Kidney Int 1999;562227- 2235
PubMedArticle
32.
Kausz  ATObrador  GTArora  PRuthazer  RLevey  ASPereira  BJG Late initiation of dialysis among women and ethnic minorities in the United States. J Am Soc Nephrol 2000;112351- 2357
PubMed
33.
Soucie  JMNeylan  JFMcClellan  W Race and sex differences in the identification of candidates for renal transplantation. Am J Kidney Dis 1992;19414- 419
PubMedArticle
34.
Kasiske  BLLondon  WEllison  MD Race and socioeconomic factors influencing early placement on the kidney transplant waiting list. J Am Soc Nephrol 1998;92142- 2147
PubMed
35.
Levey  ASGreene  TKusek  JWBeck  GJ A simplified equation to predict glomerular filtration rate from serum creatinine [abstract]. J Am Soc Nephrol 2000;11A0828
36.
Yost  KPerkins  CCohen  RMorris  CWright  W Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control 2001;12703- 711
PubMedArticle
37.
Diez-Roux  AVKiefe  CIJacobs  DR  Jr  et al.  Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies. [published correction appears in Ann Epidemiol. 2001;30:924] Ann Epidemiol 2001;11395- 405
PubMedArticle
38.
 Census 2000 Summary File 3—United States. http://factfinder.census.gov. Accessed December 22, 2005
39.
Byrne  CNedelman  JLuke  RG Race, socioeconomic status, and the development of end-stage renal disease. Am J Kidney Dis 1994;2316- 22
PubMedArticle
40.
Perneger  TVWhelton  PKKlag  MJ Race and end-stage renal disease: socioeconomic status and access to health care as mediating factors. Arch Intern Med 1995;1551201- 1208
PubMedArticle
41.
Klag  MJWhelton  PKRandall  BLMeaton  JDBrancati  FLStamler  J End-stage renal disease in African-American and White men: 16-year MRFIT findings. JAMA 1997;2771293- 1298
PubMedArticle
42.
Kirk  RE Experimental Design: Procedures for the Behavioral Sciences. 2nd ed. Belmont, Calif Brooks/Cole1982;
43.
Doescher  MPSaver  BGFranks  PFiscella  K Racial and ethnic disparities in perceptions of physician style and trust. Arch Fam Med 2000;91156- 1163
PubMedArticle
44.
Berrios-Rivera  JPStreet  RL  JrGarcia Popa-Lisseanu  MG  et al.  Trust in physicians and elements of the medical interaction in patients with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum 2006;55385- 393
PubMedArticle
45.
Owen  WF  JrSzczech  LAFrankenfield  DL Healthcare system interventions for inequality in quality: corrective action through evidence-based medicine. J Natl Med Assoc 2002;94 ((suppl)) 83S- 91S
PubMed
46.
Boulware  LETroll  MUJaar  BGMyers  DIPowe  NR Identification and referral of patients with progressive CKD: a national study. Am J Kidney Dis 2006;48192- 204
PubMedArticle
47.
Fox  CHBrooks  AZayas  LEMcClellan  WMurray  B Primary care physicians' knowledge and practice patterns in the treatment of chronic kidney disease: an Upstate New York Practice-based Research Network (UNYNET) study. J Am Board Fam Med 2006;1954- 61
PubMedArticle
48.
Abuelo  JGShemin  DChazan  JA Serum creatinine concentration at the onset of uremia: higher levels in black males. Clin Nephrol 1992;37303- 307
PubMed
49.
Cody  JDaly  CCampbell  M  et al.  Recombinant human erythropoietin for chronic renal failure anaemia in pre-dialysis patients. Cochrane Database Syst Rev 2005;3CD003266
PubMed
50.
Obrador  GTRoberts  TSt. Peter  WLFrazier  EPereira  BJCollins  AJ Trends in anemia at initiation of dialysis in the United States. Kidney Int 2001;601875- 1884
PubMedArticle
51.
Krieger  NChen  JTWaterman  PDSoobader  M-JSubramanian  SVCarson  R Choosing area based socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning: the Public Health Disparities Geocoding Project (US). J Epidemiol Community Health 2003;57186- 199
PubMedArticle
×