Do Income Level and Race Influence Survival in Patients Receiving Hemodialysis?

Article Outline

• Abstract
• Materials and Methods
  • Study Population
    • Patient Level
    • Facility Level
  • Income Level Groups
  • Measurements
    • Clinical Characteristics
    • Facility Characteristics
    • Socioeconomic Characteristics
  • End Points
  • Statistical Analyses
• Results
  • Baseline Population Characteristics
    • Clinical Characteristics
    • Dialysis Facility Characteristics
    • Socioeconomic Characteristics
  • African American versus Other Race Characteristics
    • Clinical Characteristics
    • Dialysis Facility Characteristics
    • Socioeconomic Characteristics
  • Income Group Long-Term Survival
    • Unadjusted Survival
    • Adjusted Survival
  • African American versus Other Race Long-Term Survival
    • Unadjusted Survival
    • Adjusted Survival
• Discussion
  • Dialysis Mortality and Socioeconomic Status
  • Clinical Characteristics
  • Facility Characteristics
  • Socioeconomic Characteristics
  • Race and Socioeconomic Status
• Limitations
• Conclusions
• Acknowledgment
• Appendix
• References
• Copyright

Abstract 

Background

Residence in a lower-income area has been associated with higher mortality among patients receiving dialysis. We sought to determine whether these differences persist and whether the effect of income-area on mortality is different for African Americans versus patients of other races.

Methods

We evaluated relationships between lower- and higher-income versus middle-income area residence and mortality to 5 years after adjusting for differences in baseline clinical, dialysis facility, and socioeconomic characteristics in 186,424 adult patients with end-stage renal disease initiating hemodialysis at stand-alone facilities between 1996 and 1999. We also compared mortality differences between race and income level groups using non-African Americans residing in middle-income areas as the reference group.

Results

Patients with end-stage renal disease who reside in lower-income areas were younger and more frequently African American. After adjustment, there were no mortality differences among income level groups. However, African Americans in all income level groups had lower adjusted mortality compared with the reference group (lower-income hazard ratio [HR]=0.771, 95% confidence interval [CI], 0.736-0.808; middle-income HR=0.755, 95% CI, 0.730-0.781; higher-income HR=0.809, 95% CI, 0.764-0.857), whereas adjusted mortality was similar among non–African-American income level groups (lower-income HR=1.019, 95% CI, 0.976-1.064; higher-income HR=1.003, 95% CI, 0.968-1.039).

Conclusion

Adjusted survival for patients receiving hemodialysis in all income areas was similar. However, this result masks the paradoxically higher survival for African American versus patients of other race and demonstrates the need to adjust for differences in demographic, clinical, provider, and socioeconomic status characteristics.

Keywords: Dialysis, End-stage renal disease, Income level, Race

The Centers for Medicare and Medicaid Services End-Stage Renal Disease program is the only US public program that provides health insurance coverage for a specific disease. Although financial barriers to renal replacement therapy have been eliminated, questions persist as to whether equal access to renal replacement therapy is being translated into equal health care outcomes for all patients with end-stage renal disease. These concerns are particularly relevant for patients of lower socioeconomic status and African-American patients receiving dialysis.

Previous studies have shown that lower socioeconomic status is associated with a higher incidence of end-stage renal disease,¹, ², ³ but with no difference in access to renal replacement therapy.⁴ Nonetheless, studies using information from the 1980s and early 1990s reported that residence in a lower socioeconomic status area was associated with increased mortality for patients on dialysis.⁵, ⁶, ⁷ In 1994, Medicare implemented its end-stage renal disease health care quality improvement program, which led to significant improvements in dialysis dose adequacy.⁸ Between 1993 and 2000, the proportion of patients receiving hemodialysis with a urea reduction ratio of 65% or more increased by 41% in Caucasian patients (from 46% to 87%) and by 48% in African-American patients (from 36% to 84%).⁹

We sought to determine whether survival differences between patients receiving hemodialysis residing in lower- and higher-income areas have persisted and to determine the extent to which survival differences are affected by differences in baseline patient characteristics, characteristics of the facilities within which patients receive dialysis, and characteristics of their general socioeconomic environment. We also sought to determine whether these relationships differ for African Americans versus patients of other races.

Back to Article Outline

Materials and Methods 

Study Population 

Our study used a retrospective cohort design to take advantage of the United States Renal Data System, a national database of end-stage renal disease in the United States operated by the National Institute of Diabetes and Digestive and Kidney Diseases in conjunction with the Centers for Medicare and Medicaid Services. We included incident US patients on dialysis with an initial dialysis treatment between January 1, 1996, and December 31, 1999, who survived to 90 days without censoring because of renal transplantation or loss to follow-up, with follow-up through March 31, 2001. We excluded patients aged less than 18 years at the time of dialysis initiation, patients not receiving in-center hemodialysis, and patients not receiving dialysis in a free-standing facility. US Renal Data System patient-level information was merged with 2000 US Census ZIP Code data to provide socioeconomic information for each patient's ZIP Code of residence.

US Renal Data System Annual Facility Surveys and Medicare Dialysis Facility Cost Reports were the sources for dialysis facility administrative information used in this study. These reports provided 1 set of dialysis facility administrative data for each year a dialysis facility participated in the study. Patient-level data were merged with the annual facility surveys and dialysis facility cost reports to create our combined database. Matching incident patients on dialysis with their dialysis facility on their study start date can be problematic because the facility survey covers a calendar year, but the facility cost report is not calendar-year based. We used a 2-step procedure whereby we first matched facility surveys and cost reports and then matched patient records with the combined facility file. In the first step, cost reports were matched to facility surveys. When these reports covered different time periods, the cost report was selected that covered the greater portion of the facility survey's calendar year. When matching patients with facilities, we first attempted to match the patient records with the appropriate facility record of the same year. If there was no match, we searched for alternative facility records in the following order: 1 year earlier, 1 year later, 2 years earlier, and 2 years later. If there was still no match, we attempted to match by facility surveys alone in this order: same year, 1 year earlier, and 1 year later.

Income Level Groups 

We ordered study patients by ascending median household income in their ZIP Code of residence. Patients in the lowest quartile were defined as lower income, patients in the highest quartile were defined as higher income, and patients in the middle 2 quartiles were considered middle income.

Measurements 

Candidate variables for our mortality models included baseline patient characteristics, characteristics of each patient's dialysis facility on their study start date, and socioeconomic characteristics of patients and patients' ZIP Code of residence in 2000.

Candidate patient characteristics obtained from the Centers for Medicare and Medicaid Services Medical Evidence Report at initial patient registration included demographic, clinical history, and laboratory values obtained before the first dialysis treatment. Demographic measurements were birth date, sex, race, height, weight; clinical history measurements were diabetes cause of end-stage renal disease (primary or contributing), comorbid conditions (congestive heart failure, ischemic heart disease, myocardial infarction, cardiac arrest, cardiac dysrhythmia, pericarditis, cerebrovascular disease, history of hypertension, diabetes [currently on insulin], chronic obstructive pulmonary disease, tobacco use [current smoker], malignant neoplasm [cancer], alcohol dependence, drug dependence, human immunodeficiency virus positive status, acquired immunodeficiency syndrome, inability to ambulate, and inability to transfer), and predialysis erythropoietin (EPO) administration; and laboratory values included hematocrit (%), hemoglobin (grams/deciliter), serum albumin (grams/deciliter), serum creatinine (milligrams/deciliter), and blood urea nitrogen (milligrams/deciliter). In addition, we calculated body mass index in kilometer/meters squared in patients for whom height and weigh data were available, and we calculated patient ages as of their study start dates.

Candidate dialysis facility characteristics included for-profit corporate ownership, number of patients on hemodialysis per clinic, average number of treatments per patient per week, in-home dialysis provision, and total number of nurses. We also calculated 2 operating ratios (patient-to-nurse and patient-to-total staff), and the percentage of registered nurses in the total nursing staff.

Candidate socioeconomic characteristics at the patient level were obtained from the Medical Evidence Report at initial patient registration. These data include type of insurance coverage, employment status, and US census region of residence. Residential ZIP Code characteristics that were used as candidate variables in our models include the percentage of African-American population, median household income, average house value, and rural/urban metropolitan statistical area.

End Points 

Our primary analysis assessed the relationships of income level groups with survival of patients on in-center hemodialysis to 5 years. Secondary analyses compared these relationships between African Americans and patients of other races.

Statistical Analyses 

Baseline patient, facility, and socioeconomic characteristics are presented as percentages for discrete variables and as medians (25th and 75th percentiles) for continuous variables. Kaplan–Meier estimates were generated by income level group. By using the middle-income group as a reference, adjusted hazard ratios (HRs) (with 95% confidence intervals [CIs]) were determined using Cox proportional-hazards modeling techniques with adjustment for patients on hemodialysis clustering around dialysis facilities and with censoring on kidney transplant or loss to follow-up. Nonlinear continuous variables were transformed using restricted cubic spline techniques.

Four adjusted models were developed. The first model included only income level group, age, and gender. For the second model, a combination of stepwise and backward selection methods were used to select clinical and demographic variables that were used in this and subsequent models. The third model included dialysis facility characteristics along with income level group and baseline clinical characteristics. The fourth model included socioeconomic characteristics along with income level group, baseline clinical characteristics, and baseline dialysis facility characteristics. Proportional hazards and linearity assumptions were checked. Violations of linearity assumptions were resolved using variable transformations.

We performed additional analyses to test whether income level effects on survival of patients on hemodialysis were similar in African Americans versus patients of other races. Specifically, we repeated our previously described baseline characteristics, Kaplan–Meier modeling, and Cox proportional-hazards analyses comparing mortality for different race (African American or other race) and income level groups with middle-income non-African Americans as the reference.

Back to Article Outline

Results 

Baseline Population Characteristics 

Median household income ranges for our 3 strata were lower income (<$32,043), middle-income ($32,043-$44,246), and higher income (>$44,246). Application of these criteria divided our population into 46,457 lower-income, 93,212 middle-income, and 46,370 higher-income patients on hemodialysis. Lower income was associated with younger age, female gender, and African-American race (Table 1). Lower-income individuals were more likely to be hypertensive and to have diabetes as their primary cause of end-stage renal disease. Higher income was associated with greater predialysis EPO use and ischemic heart disease. However, there were few differences in other clinical characteristics, such as laboratory values, body mass index, or micro- and macrovascular disease.

Table 1. Patient Characteristics by Income Stratum

Characteristic (% unless otherwise noted)	Income≤Q1	Q1<Income<Q3	Q3≥Income	P Value
n	46,457	93,212	46,370
Clinical
Age in years: median (Q1, Q3)	62(50,72)	64(52,74)	66(54,75)	<.001
Sex (% female)	50.0	47.7	45.0	<.001
Race (% African American)	47.7	33.9	19.1	<.001
Body mass index: median (Q1, Q3)	25.5(22.1,30.1)	25.3(21.97,29.8)	24.9(21.8,29.1)	<.001
Diabetes (primary or contributing)	44.6	41.9	40.3	<.001
Predialysis EPO administration	19.6	23.8	29.8	<.001
Congestive heart failure	31.5	33.3	32.2	<.001
Ischemic heart disease	19.1	23.2	24.1	<.001
Myocardial infarction	6.6	8.6	9.2	<.001
Cardiac arrest	0.8	0.7	0.7	.744
Cardiac dysrhythmia	4.0	5.3	5.9	<.001
Pericarditis	0.9	0.9	1.0	.052
Cerebrovascular disease	8.6	9.1	8.6	.002
Peripheral vascular disease	12.8	14.3	14.2	<.001
History of hypertension	78.3	76.4	72.4	<.001
Diabetes (currently on insulin)	24.2	24.3	22.1	<.001
Chronic obstructive pulmonary disease	5.9	7.2	6.6	<.001
Tobacco use (current smoker)	6.1	5.8	3.9	<.001
Malignant neoplasm (cancer)	3.8	5.1	5.3	<.001
Drug dependence	1.4	1.1	0.7	<.001
Human immunodeficiency virus, positive	1.1	0.8	0.5	<.001
Inability to ambulate	4.1	3.7	3.4	<.001
Inability to transfer	1.4	1.2	1.0	<.001
Hematocrit: median (Q1, Q3)	27.7(24.3,31.2)	28.4(25.0,31.8)	28.8(25.5,32.2)	<.001
Hemoglobin (g/dL): median (Q1, Q3)	9.2(8.1,10.4)	9.40(8.3,10.6)	9.60(8.5,10.7)	<.001
Serum albumin (g/dL): median (Q1, Q3)	3.2(2.7,3.6)	3.20(2.8,3.6)	3.20(2.8,3.7)	<.001
Serum creatinine (mg/dL): median (Q1, Q3)	7.7(5.8,10.1)	7.3(5.6,9.5)	7.3(5.6,9.3)	<.001
Blood urea nitrogen (mg/dL): median (Q1, Q3)	85.0(66.0,108.0)	86.0(66.0,109.0)	88.0(68.0,111.0)	<.001
Facility
Ownership, for-profit corporation	70.0	64.9	64.7	<.001
Hemodialysis: patients per clinic: median (Q1, Q3)	79(51,116)	83(56,119)	81(57,113)	<.001
No. of treatments per patient/wk (n): median (Q1, Q3)	3.0(3.0,3.0)	3.0(3.0,3.0)	3.0(3.0,3.0)	<.001
In-home dialysis provision	51.3	57.9	55.9	<.001
Total nurses	9.1	9.3	8.4	<.001
Percent RN of total nursing staff	76.8	81.0	83.1	<.001
Patient-to-nurse ratio	11.5	12.0	13.3	<.001
Patient-to-total-staff ratio	4.8	4.6	4.8	<.001
Socioeconomic
Insurance coverage				<.001
Medicare+Medicaid	17.9	12.1	9.2
Medicaid	17.0	12.5	9.0
Medicare	35.4	41.8	44.0
Other	29.7	33.6	37.8
Employment status				<.001
Unemployed	26.5	20.8	15.8
Employed	13.1	14.7	17.9
Other	60.4	64.6	66.3
ZIP Code demographics
African-American population (%)	31.3	21.5	12.1	<.001
Median household income ($): median (Q1, Q3)	27,471(22,917;31,690)	36,234(30,909;42,594)	50,744(42,042;61,769)	<.001
Average house value ($): median (Q1, Q3)	61,000(49,100;78,900)	89,700(71,300;120,200)	157,600(118,200;213,300)	<.001
Rural MSA	36.7	15.6	3.9	<.001

RN=Registered Nurse; MSA=metropolitan statistical area; EPO=erythropoietin.

Lower-income individuals dialyzed in somewhat smaller facilities that more often had for-profit corporate ownership and less frequently offered in-home dialysis as a treatment option (Table 1). However, lower-income individuals also dialyzed in facilities with lower patient-to-nurse ratios. Otherwise, dialysis facility characteristics were remarkably similar across all 3 income strata.

Lower income was associated with greater Medicaid insurance coverage (alone or with Medicare), greater unemployment, and a greater likelihood of residency in a rural area with a higher percentage of African Americans (Table 1). Higher-income individuals tended to live in areas with higher valued houses.

African American versus Other Race Characteristics 

Across all income strata, African-American patients were younger, with greater hypertension, drug dependence, and human immunodeficiency virus positive status (Table 2). Other race patients were more often female, with greater overall micro- and macrovascular disease and cancer, but more predialysis EPO use. Predialysis EPO use increased with income level in both racial groups.

Table 2. Clinical Characteristics by Income Level and Race

	African American				Other Race
Characteristic(%)	Q1	Q2	Q3	P Value	Q1	Q2	Q3	P Value
n	22,163	31,633	8860		24,294	61,579	37,510
Clinical
Age in years: median (Q1, Q3)	60(47,70)	59(47,69)	59(47,69)	.002	65(53,73)	67(55,75)	68(56,76)	<.001
Sex (% female)	45.9	47.1	50.3	<.001	53.7	55.1	56.1	<.001
Body mass index: median (Q1, Q3)	26.0(22.2,30.9)	26.0(22.3,31.0)	25.9(22.2,30.7)	.086	25.1(21.9,29.3)	25.0(21.8,29.4)	24.7(21.6,28.7)	<.001
Diabetes primary or contributing	41.0	41.6	40.4	.109	47.8	42.0	40.3	<.001
Predialysis EPO administration	17.4	21.5	25.8	<.001	21.6	24.9	30.7	<.001
Congestive heart failure	28.2	26.7	26.2	<.001	34.4	35.6	33.6	<.001
Ischemic heart disease	12.8	14.3	14.8	<.001	24.9	27.8	26.3	<.001
Myocardial infarction	4.5	5.3	5.2	<.001	8.5	10.3	10.1	<.001
Cardiac arrest				.594				.023
Cardiac dysrhythmia	2.8	3.1	3.1	.141	5.1	6.4	6.6	<.001
Pericarditis	0.9	0.9	1.0	.768	1.0	0.9	1.0	.030
Cerebrovascular disease	8.5	8.7	7.9	.042	8.7	9.2	8.7	.008
Peripheral vascular disease	9.1	9.6	9.5	.147	16.2	16.7	15.3	<.001
History of hypertension	81.1	81.4	75.9	<.001	75.7	73.8	71.6	<.001
Diabetes (currently on insulin)	22.3	23.0	20.9	<.001	26.0	24.9	22.4	<.001
Chronic obstructive pulmonary disease	3.6	3.9	3.6	.196	7.9	8.9	7.3	<.001
Tobacco use (current smoker)	6.4	5.8	4.0	<.001	5.8	5.8	3.9	<.001
Malignant neoplasm (cancer)	3.1	3.6	3.6	.001	4.5	5.9	5.7	<.001
Drug dependence	2.5	2.4	2.1	.131	0.5	0.4	0.3	.014
HIV positive status	2.2	2.0	2.2	.272	0.2	0.1	0.1	.004
Inability to ambulate	3.9	3.5	3.0	.001	4.4	3.8	3.5	<.001
Inability to transfer	1.3	1.2	0.9	.007	1.4	1.2	1.0	<.001
Hematocrit: median (Q1, Q3)	27.0(23.6,30.7)	27.5(24.1,31.0)	27.9(24.5,31.3)	<.001	28.2(25.0,31.7)	28.8(25.6,32.1)	29.0(25.8,32.4)	<.001
Hemoglobin (g/dL): median (Q1, Q3)	8.9(7.8,10.1)	9.1(8.0,10.3)	9.2(8.1,10.4)	<.001	9.4(8.4,10.6)	9.6(8.6,10.7)	9.7(8.6,10.8)	<.001
Serum albumin (g/dL): median (Q1, Q3)	3.2(2.7,3.6)	3.2(2.7,3.6)	3.2(2.7,3.6)	.001	3.2(2.7,3.6)	3.2(2.8,3.6)	3.2(2.8,3.7)	<.001
Serum creatinine (mg/dL): median (Q1, Q3)	8.5(6.5,11.1)	8.3(6.4,10.8)	8.4(6.4,10.8)	<.001	7.0(5.3,9.1)	6.9(5.3,8.8)	7.0(5.4,9.0)	<.001
Creatinine clearance (ml/min): median (Q1, Q3)	0.0(0.0,0.0)	0.0(0.0,3.0)	0.0(0.0,0.0)	<.001	0.0(0.0,0.0)	0.0(0.0,6.8)	0.0(0.0,5.0)	<.001
Blood urea nitrogen (mg/dL): median (Q1, Q3)	85.0(65.0,108.0)	84.0(65.0,107.0)	84.0(66.0,107.0)	.021	86.0(66.0,107.0)	87.0(67.0,110.0)	89.0(69.0,112.0)	<.001
Facility
Ownership, for-profit corporation	70.5	67.7	69.4	<.001	69.6	63.5	63.6	<.001
Patients on hemodialysis per clinic: median (Q1, Q3)	86(58,125)	91(63,133)	88(60,126)	<.001	73(45,112)	78(53,113)	80(57,111)	<.001
No. of treatments per patient per week (n): median (Q1, Q3)	3.0(3.0,3.0)	3.0(3.0,3.0)	3.0(3.0,3.0)	<.001	3.0(3.0,3.0)	3.0(3.0,3.0)	3.0(3.0,3.0)	.003
Facility age (years)	10.1	10.0	9.4	<.001	9.2	9.4	9.3	<.001
In-home dialysis provision	50.2	57.3	53.6	<.001	52.4	58.2	56.4	<.001
Total nurses	9.1	9.8	8.8	<.001	9.0	9.1	8.2	<.001
Percent RN of total nursing staff	77.1	80.0	82.9	<.001	76.6	81.5	83.2	<.001
Patient-to-nurse ratio	12.1	12.5	13.6	.007	11.0	11.7	13.3	<.001
Patient-to-total-staff ratio	4.8	4.7	4.9	<.001	4.7	4.6	4.8	<.001
Socioeconomic
Insurance				<.001				<.001
Medicare+Medicaid	20.1	1537	12.1		16.0	10.3	8.5
Medicaid	20.3	16.8	15.0		14.1	10.3	7.6
Medicare	28.4	29.8	28.5		41.8	47.9	47.7
Other	31.3	37.6	44.5		28.2	31.5	36.2
Employment				<.001				<.001
Unemployed	33.2	28.1	24.0		20.44	17.0	13.8
Employed	11.0	13.2	17.4		15.1	15.4	18.0
Other	55.8	58.7	58.6		64.5	67.6	68.2
ZIP Code demographics
African-American population (%)	53.6	44.5	34.0	<.001	10.9	9.7	7.0	<.001
Median household income ($): median (Q1, Q3)	26,723(22,006;30,954)	32,968(27,317;39,253)	45,960(36,502; 56,973)	<.001	28,468(23,833; 32,333)	37,714(32,494;44,322)	51,736(43,265;62,895)	<.001
Average house value ($)	59,900(49,200;77,100)	82,200(64,300;109,000)	136,100(97,400;178,800)	<.001	61,700(48,800; 80,300)	93,500(75,600;126,100)	162,600(123,900;220,100)	<.001
Rural MSA	31.0	10.3	2.7	<.001	42.0	18.4	4.2	<.001

HIV=human immunodeficiency virus; RN=Registered Nurse; MSA=metropolitan statistical area.

African Americans tended to dialyze in larger facilities that required more human and other resources to operate (Table 2). Otherwise, there were no race-based differences in key operating indicators within income level groups.

Higher-income individuals in both racial groups tended to be urban and reside in areas with more expensive houses. Within each income-level group, African Americans were less likely to reside in rural areas and more likely to reside in areas with lower household incomes and lower valued houses (Table 2). African Americans also made more use of Medicaid insurance and were more often unemployed at the time of dialysis initiation.

Income Group Long-Term Survival 

Unadjusted survival for lower-income individuals was greater than that for higher-income individuals throughout the follow-up period, with the modest survival advantage for lower-income versus higher-income individuals at 1 year (79.9% vs 79.1%) increasing to 4.7% at 4 years follow-up (41.6% vs 36.9%) (Table 3).

Table 3. Unadjusted Survival by Income Level and Race/Income Level

	Survival of Patients (%) on Dialysis at:
	1 y	2 y	3 y	4 y
All Patients
Income≤Q1	79.9	65.1	52.7	41.6
Q1<Income<Q3	79.3	63.9	50.8	38.7
Q3≥Income	79.1	63.3	49.9	36.9
African-American Patients
Income≤Q1	83.2	70.5	59.3	48.0
Q1<Income<Q3	84.3	72.4	60.8	49.2
Q3≥Income	84.2	72.6	60.6	48.2
Patients of Other Race
Income≤Q1	76.8	60.0	46.2	35.2
Q1<Income<Q3	76.7	59.4	45.3	32.6
Q3≥Income	77.9	61.0	47.2	34.0

Figure 1 shows adjusted HRs for mortality in the lower- and higher-versus middle-income groups. Adjustment for clinical characteristics (including race) reversed the unadjusted risk for mortality so that lower income became associated with increased mortality (HR=1.056, 95% CI, 1.023-1.091) and higher income became associated with reduced mortality (HR=0.958, 95% CI, 0.932-0.985). There was little change after adjustment for both facility and clinical characteristics. Further adjustment for socioeconomic characteristics (total of 42 adjustment variables) eliminated all survival differences between individuals in the 3 income-level groups such that mortality for patients residing in lower- versus middle-income areas (HR=1.018, 95% CI, 0.984-1.054) was similar to that of patients residing in higher-income areas (HR=1.014, 95% CI, 0.980-1.049). Patient characteristics most associated with increased survival in descending chi-square value order included younger age, lower glomerular filtration rate, higher body mass index, and higher serum albumin at study entry (Appendix 1).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  HRs for the lower- and upper-income stratum versus middle-income before and after adjustment. SES = socioeconomic status.

African American versus Other Race Long-Term Survival 

Four-year unadjusted survival for African Americans was higher than that for other races in all income level groups (lower income, 48.0% vs 35.2%; middle income, 49.2% vs 32.6%; higher income, 48.2% vs 34.0%) (Table 3).

Figure 2 shows HRs comparing mortality for patients in the reference group (middle income, other race) with patients in different race and income level groups. After adjustment for age and gender, all African American patients and patients of other race in the higher-income group have lower mortality than patients in the reference group, whereas lower-income patients of other race have higher mortality. Further adjustment for clinical and facility characteristics made little change. However, additional adjustment for socioeconomic characteristics (total of 44 adjustment variables) eliminated mortality differences among other race patients while maintaining or increasing the survival advantage for African Americans in all income areas (lower-income HR=0.771, 95% CI, 0.736-0.808; middle-income HR=0.755, 95% CI, 0.730-0.781; higher-income HR=0.809, 95% CI, 0.764-0.857). Patient characteristics most associated with increased survival in descending chi-square value order included younger age, lower glomerular filtration rate, higher body mass index, and higher serum albumin at study entry (Appendix 1).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  HRs by race and income group compared with other race middle-income group. SES=socioeconomic status.

Back to Article Outline

Discussion 

Dialysis Mortality and Socioeconomic Status 

Previous population-based studies in patients on dialysis found general relationships between lower socioeconomic status and higher mortality, but indicated that this relationship might be limited to African Americans.⁵, ¹⁰ These studies differed in population, time period of analysis, and adjustment variables. By using Michigan patients initiating dialysis between 1980 and 1987, Port and colleagues⁵ found a significant inverse relationship between the adjusted mortality risk for African Americans and the average household income in their residential area; the trend for Caucasians was not significant. Garg et al¹⁰ confirmed the relationship between residential area income and patient mortality using a small population with end-stage renal disease (n=3165) from the early 1990s. Our work expands on these earlier studies by demonstrating that an inverse relationship, between mortality adjusted for clinical factors and residential area income, persists. We believe that this relationship is in part an artifact of the distribution of African-American patients across the 3 income levels; as income in area of residence increases, the percentage of African Americans decreases.

Clinical Characteristics 

The key differences among groups were age (median 62 years for lower income and 66 years for higher income) and African-American race (47.7% for lower income and 19.1% for higher income). In subgroup analyses we found median age at dialysis initiation for African Americans varied between 59 and 60 years in the 3 income levels, whereas median age for individuals of other races varied between 65 and 68 years. Thus, it seems that the observed relationship between baseline age and income level was driven by the racial composition of the 3 income level groups.

Facility Characteristics 

Facility size (number of patients on hemodialysis per clinic) was a significant predictor of mortality in our fully adjusted model. However, facility size differences between the 2 racial groups were modest. In all models, the HRs after adjusting for clinical and facility characteristics were similar to those after adjusting for clinical characteristics alone. Thus, it seems that the impact of facility characteristics is not associated with income level or racial differences among patients on hemodialysis.

Socioeconomic Characteristics 

Several socioeconomic status characteristics were associated with mortality differences in our fully adjusted model. Medicare and Medicaid insurance (alone or together) and a greater percentage of African Americans in a patient's residential area were associated with higher mortality, whereas employment (full or part-time) and rural versus urban residency were associated with lower mortality. All of these variables are influenced in part by the income level of a patient's residential area. However, the association between rural residency and lower mortality observed in this study is opposed to what one would expect from income level information alone. Thus, it may well be an artifact of other socioeconomic status variables in our model.

Race and Socioeconomic Status 

End-stage renal disease treatment for African Americans in all income levels begins at 5 to 8 years younger age than for patients of other races; yet, African-American patients on hemodialysis have greater survival (unadjusted and adjusted) than patients of other races. Further, there seems to be a relationship between race and income level such that although socioeconomic factors lower the survival prospects for lower-income patients in both racial groups, only patients in the other race group reap survival benefit from higher income. This observation lends support to the “diminishing returns hypothesis” for African Americans.¹¹

Back to Article Outline

Limitations 

This study was an observational analysis conducted in a database collected for other purposes. As such, we cannot exclude that our findings may be subject to the influence of unmeasured confounders. For example, previous studies have shown relationships between compliance and various socioeconomic factors of patients with end-stage renal disease.¹², ¹³, ¹⁴ However, this is not a plausible explanation because African-American race is a risk factor for lower mortality across all income levels. Similarly, our results could be attributable to differences in dialysis dose for African Americans; however, this would be unexpected because previous researchers have found that the adequacy of dialysis has improved for African Americans and is now comparable to the adequacy of dialysis for other races.⁹, ¹⁵

Back to Article Outline

Conclusions 

Medicare's end-stage renal disease program seems to have eliminated previously described income-based mortality differences among patients on dialysis, resulting in similar survival for patients receiving hemodialysis who reside in lower-, middle-, and higher-income areas. However, this result masks the paradoxically higher survival for African-American patients versus patients of other race across income areas and demonstrates the need to adjust for differences in demographic, clinical, provider, and socioeconomic status characteristics.

Back to Article Outline

Acknowledgment 

The authors thank Maqui Ortiz for expert editorial assistance and article preparation.

Back to Article Outline

Appendix 

Appendix. Two Fully Adjusted Analyses Were Developed for This Study. We Have Included the Details for Both Models in This Section.

Back to Article Outline

References 

1. Young EW, Mauger EA, Jiang KH, et al. Socioeconomic status and end-stage renal disease in the United States. Kidney Int. 1994;45:907–911
   • View In Article
   • MEDLINE
   • CrossRef
2. Perneger TV, Klag MJ, Whelton PK. Race and socioeconomic status in hypertension and renal disease. Curr Opin Nephrol Hypertens. 1995;4:235–239
   • View In Article
   • MEDLINE
   • CrossRef
3. Klag MJ, Whelton PK, Randall BL, et al. End-stage renal disease in African-American and white men (16-year MRFIT findings). JAMA. 1997;277:1293–1298
   • View In Article
   • MEDLINE
4. Powe NR, Tarver-Carr ME, Eberhardt MS, Brancati FL. Receipt of renal replacement therapy in the United States: a population-based study of sociodemographic disparities from the Second National Health and Nutrition Examination Survey (NHANES II). Am J Kidney Dis. 2003;42:249–255
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (94 KB)
   • CrossRef
5. Port FK, Wolfe RA, Levin NW, et al. Income and survival in chronic dialysis patients. ASAIO Trans. 1990;36:M154–M157
   • View In Article
   • MEDLINE
6. Garg PP, Frick KD, Diener-West M, Powe NR. Effect of the ownership of dialysis facilities on patients' survival and referral for transplantation. N Engl J Med. 1999;341:1653–1660
   • View In Article
   • MEDLINE
   • CrossRef
7. Devereaux PJ, Schunemann HJ, Ravindran N, et al. Comparison of mortality between private for-profit and private not-for-profit hemodialysis centers: a systematic review and meta-analysis. JAMA. 2002;288:2449–2457
   • View In Article
   • MEDLINE
   • CrossRef
8. McClellan WM, Frankenfield DL, Frederick PR, et al. Improving the care of ESRD patients: a success story. Health Care Financ Rev. 2003;24:89–100
   • View In Article
   • MEDLINE
9. Sehgal AR. Impact of quality improvement efforts on race and sex disparities in hemodialysis. JAMA. 2003;289:996–1000
   • View In Article
   • MEDLINE
   • CrossRef
10. Garg PP, Diener-West M, Powe NR. Income-based disparities in outcomes for patients with chronic kidney disease. Semin Nephrol. 2001;21:377–385
    • View In Article
    • Abstract
11. Farmer MM, Ferraro KF. Are racial disparities in health conditional on socioeconomic status?. Soc Sci Med. 2005;60:191–204
    • View In Article
    • MEDLINE
    • CrossRef
12. Loghman-Adham M. Medication noncompliance in patients with chronic disease: issues in dialysis and renal transplantation. Am J Manag Care. 2003;9:155–171
    • View In Article
    • MEDLINE
13. Rovelli M, Palmeri D, Vossler E, et al. Noncompliance in organ transplant recipients. Transplant Proc. 1989;21:833–834
    • View In Article
    • MEDLINE
14. Swanson MA, Palmeri D, Vossler ED, et al. Noncompliance in organ transplant recipients. Pharmacotherapy. 1991;11:173S–174S
    • View In Article
    • MEDLINE
15. Owen WF, Szczech LA, Frankenfield DL. Healthcare system interventions for inequality in quality: corrective action through evidence-based medicine. J Natl Med Assoc. 2002;94:83S–91S
    • View In Article
    • MEDLINE