Survival Advantage of Hispanic Patients Initiating Dialysis in the ...

Survival Advantage of Hispanic Patients Initiating Dialysis in the United States Is Modified by Race Bhamidipati V.R. Murthy,* Donald A. Molony,* and Austin G. Stack*† *Division of Renal Diseases and Hypertension, Department of Internal Medicine, University of Texas Medical School at Houston, Houston, Texas; and †Regional Kidney Center, Department of Internal Medicine, Letterkenny General Hospital, North Western Health Board, Letterkenny, Ireland Differences in survival have been reported among ethnic groups in the general population. Whether these extend to patients with ESRD is unclear. Using national data, mortality risks of ethnic groups who began dialysis treatment in the United States between May 1, 1995, and July 31, 1997, were compared over 2 yr. Patients were classified as Hispanic or non-Hispanic and then subclassified by race forming six race-specific subgroups: Hispanic white, black, and other and non-Hispanic white, black, and other. Mortality rates for Hispanics compared with non-Hispanics were 19.2 versus 26 per 100 patient-years at risk for those with diabetes and were 14.7 versus 22.7 per 100 patient-years at risk for those without diabetes. For those with diabetes, adjusted mortality risks for Hispanics versus non-Hispanics were 30% lower (95% confidence interval [CI], 26 to 34%). In subgroup analysis, mortality risks for Hispanic whites and Hispanic blacks were 35% (95% CI, 31 to 39%) and 33% (95% CI, 12 to 48%) lower than non-Hispanic whites and were similar in magnitude to those of non-Hispanic blacks (32% lower; 95% CI, 29 to 35%) and non-Hispanic other (33% lower; 95% CI, 28 to 39%). Interestingly, mortality risks for Hispanic others were not significantly different from non-Hispanic whites. For those without diabetes, adjusted mortality risks for Hispanics versus non-Hispanics were 17% lower (95% CI, 9 to 23%), and subgroup analysis yielded similar patterns to those of individuals with diabetes. The survival advantage of Hispanic over non-Hispanic patients who receive chronic dialysis treatment in the United States is not consistent across subgroups and is modified by race. Cultural and genetic differences as well as variation in the access and delivery of care before and while on dialysis may account for these differences. J Am Soc Nephrol 16: 782–790, 2005. doi: 10.1681/ASN.2004080627

H

ealth outcomes for ethnic minorities have become an increasingly important area of concern for health care providers and public policy leaders in the United States. This is due, in part, to the increasing number of reports that have documented significantly greater comorbidity risk profiles for Hispanics and other ethnic minorities compared with white populations (1). In addition, Hispanics have become the fastest growing ethnic minority in the general US population (2), as well as the fastest growing ethnic group among the adult ESRD population (3). Accordingly, it is important that we assess the utilization of health care services among these groups, address any disparities that exist in the receipt of such services, and evaluate any potentially negative consequences on morbidity and mortality. There is a relative paucity of data on health-related outcomes among Hispanics with chronic kidney disease in the United

Received August 1, 2004. Accepted December 22, 2004. Published online ahead of print. Publication date available at www.jasn.org. Address correspondence to: Dr. Bhamidipati V.R. Murthy, Department of Internal Medicine, Division of Renal Diseases and Hypertension, University of Texas Health Science Center at Houston, 6431 Fannin Street, MSB 4.148, Houston, TX 77030. Phone: 713-500-6868; Fax: 713-500-6882; E-mail: [email protected] The data reported here were supplied by the United States Renal Data System. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as the official policy or interpretation of the United States government. Copyright © 2005 by the American Society of Nephrology

States. A recent report by Frankenfield et al. (4) demonstrated greater survival of adult Hispanic ESRD patients who were treated with hemodialysis (HD) compared with non-Hispanic whites. Moreover, in their study, the delivery of adequate ESRD care, as measured by an achieved solute clearance (Kt/V) of ⱖ1.2, an average hemoglobin ⱖ11g/dl, or the proportion of patients with an arteriovenous fistula, was at least as good for Hispanic as for non-Hispanic white patients (4). These observations suggested that the health care provided to adult Hispanic patients who were treated with HD and associated survival experiences were similar if not better than other racial or ethnic groups. Contrasting, Kausz et al. (5) observed that Hispanics and other minorities receive comparatively less care before ESRD initiation compared with whites. In a nationally representative cohort of new ESRD patients in the United States, they observed that Hispanic patients as well as other ethnic minority groups were less likely to be initiated on dialysis at the recommended level of GFR when compared with non-Hispanic white patients. Their findings suggest that disparities exist in the provision of health care during the pre-ESRD period, and these health care disparities may have an impact on subsequent survival. In view of these conflicting observations, we explored survival differences between Hispanic and non-Hispanic subgroups among incident dialysis patients in the United States, in the context of differences in measurable baseline comorbidity and delivered pre-ESRD care. ISSN: 1046-6673/1603-0782

J Am Soc Nephrol 16: 782–790, 2005

Materials and Methods Study Design and Data The study was a historical prospective cohort of new ESRD patients in the United States who were initiated on dialysis between May 1, 1995, and July 31, 1997. Baseline data on all patients at study entry were obtained from the Center for Medicare and Medicaid Services (CMS) Medical Evidence Form 2728. The CMS form is a government document that is completed for all new patients who are initiated on dialysis (6). It is a validated data collection instrument (7), and data arising from it have been published in several previous studies (5,8 –10). The form records data on several sociodemographic characteristics and captures information on a broad list of comorbid conditions, laboratory values, and indicators of pre-ESRD care recorded at or near ESRD onset.

Definition of Ethnicity and Race The CMS form incorporates two questions that collect information on ethnicity and race. The first classifies patients as Hispanic-Mexican, Hispanic other, and non-Hispanic. The second categorizes patients into one of nine racial groups: white, black, American Indian/Alaskan Native, Asian, Pacific Islander, Mid-East/Arabian, Indian subcontinent, other, and unknown. For the purposes of this study, ethnicity was classified as either Hispanic or non-Hispanic and race was classified in one of three principal groups: white, black, and other. The “other” category comprised a relatively small fraction of the entire cohort and was included to facilitate comparisons. The study start date for all incident patients was defined as day 90 of ESRD. This was based on the fact that many patients who are younger than 65 yr do not become eligible for Medicare for up to 90 d and therefore may have incomplete claims data before this. For the current study, the Medical Evidence data set was merged with mortality and transplantation data from the US Renal Data System Standard Analysis Files. Approval for the current study was obtained from the University of Texas Institutional Review Board.

Patient Population There were 158,685 patients, aged 18 and over, who were initiated on dialysis from May 1, 1995, to July 31, 1997. Patients were excluded from the analysis if they had received a renal transplant within the first 90 d of ESRD initiation, when modality assignment could not be determined at day 90 of ESRD, or when they had died within the first 90 d (n ⫽ 19,648). An additional 38,419 patients were excluded because of missing data on demographic or clinical variables. After these exclusions, 100,618 adult patients were available for this analysis.

Statistical Analyses As survival differences have been reported among racial subgroups of ESRD patients (11,12), we believed that it was appropriate to subclassify ethnicity by race. Descriptive statistics of patient characteristics by Hispanic status were performed for the entire cohort and for each of six combined ethnic-racial subgroups: Hispanic white, Hispanic black, Hispanic other, non-Hispanic white, non Hispanic black, and nonHispanic other. Categorical variables among groups were compared using ␹2 tests, and means of continuous variables were compared using ANOVA. The health status of patients at ESRD onset and measures of preESRD care were compared between racial/ethnic subgroups. Multivariate logistic regression was used to explore the association of ethnicity and race (six subgroups) with three routinely measured comorbid indicators as well as a measure of pre-ESRD care. Comorbid indicators were represented by diabetes, coronary artery disease, anemia, and hypoalbuminemia as recorded at ESRD initiation. Severe anemia was

Survival of Hispanic Patients on Dialysis

783

defined as a hematocrit ⬍28%, and hypoalbuminemia was defined as a serum albumin ⬍3.5g/dl. These cutoff values were chosen on the basis of their threshold prognostic significance in recent mortality analyses. Furthermore, a hematocrit value of 28% represented the mean for the entire cohort, and a serum albumin value of 3.5g/dl was virtually identical to the mean of all lower limits of normal albumin values recorded on the CMS 2728 form. In addition, the use of erythropoietin before ESRD onset, also captured by the CMS document, served as a surrogate measure of health care during this period. Unadjusted and adjusted odds ratios (OR) were determined in each case with corresponding 95% confidence intervals (CI). Multivariate Cox regression models explored the relationship of ethnicity-race with mortality risk in new ESRD patients. Patients were followed until death, loss to follow-up, transplantation, or the end of 2 yr from the study start date. Both unadjusted and adjusted hazard ratios were calculated for each ethnic subgroup: Hispanic white, Hispanic black, Hispanic other, non-Hispanic black, and non-Hispanic other, with non-Hispanic white as the referent group. The following variables were adjusted for in the multivariate analyses: age, gender, hypertension, coronary artery disease, cerebrovascular disease, peripheral vascular disease, tobacco use, cardiac arrest or arrhythmia, chronic obstructive pulmonary disease, cancer, alcohol dependence, drug dependence, AIDS, functional status (inability to ambulate or transfer), body mass index (BMI), serum albumin, and hematocrit. Age was modeled as a continuous variable and all others as categorical variables. GFR at initiation of dialysis was estimated using the modified Modification of Diet in Renal Disease equation (13). Diabetic and nondiabetic patients were analyzed in separate models on finding a significant interaction term demonstrating nonproportional hazards. All statistical analyses were carried out using SAS software (V8.0; Cary, NC).

Results Baseline Differences between Hispanic and Non-Hispanic Patients Who Were Initiated on Dialysis The study cohort consisted of 100,618 patients who were initiated on dialysis between May 1, 1995 and July 31, 1997. Of these, 10,393 (10.3%) were identified as Hispanic. Table 1 illustrates differences in baseline characteristics between Hispanic and non-Hispanic patients. Compared with non-Hispanic ESRD patients, Hispanics were younger (59.2 ⫾ 15.0 versus 61.8 ⫾ 15.3 yr), had larger proportion of whites (87.1 versus 60.7%), and had a higher percentage of patients with diabetic nephropathy (61.3 versus 42.3%). The prevalence of diabetes (either as a cause of ESRD or comorbid condition), hypertension, tobacco use, coronary artery disease or myocardial infarction, congestive heart failure, chronic lung disease, cerebral and peripheral vascular disease, AIDS, and cancer were significantly lower in Hispanic patients compared with non-Hispanic patients. Although serum creatinine values were virtually identical in both groups, the estimated GFR values at ESRD initiation was significantly lower for Hispanic patients compared with non-Hispanic patients (6.8 ⫾ 2.7 versus 7.1 ⫾ 2.8 ml/min). Mean serum albumin (3.1 ⫾ 0.7 versus 3.2 ⫾ 0.7 g/dl) and hematocrit levels (27.6 versus 28.1%) were also significantly lower in the Hispanic group. The percentage of patients who were treated with peritoneal dialysis was significantly lower in Hispanic compared with non-Hispanic patients (10.4 versus 13.3%). The demographic and clinical characteristics of patients who were excluded from this analysis as a result of missing or

784

Journal of the American Society of Nephrology

J Am Soc Nephrol 16: 782–790, 2005

Table 1. Differences in baseline variables between Hispanics and non-Hispanics who were initiated on dialysis in the United States Variable

Age at ESRD onset (yr) Gender Male Female Race White Black Other Diabetes (cause of ESRD) Comorbid conditions (% present) Diabetes (cause of ESRD or comorbidity) Hypertension Tobacco use (current or past) Chronic obstructive lung disease Coronary artery diseasec Myocardial infarction Congestive heart failure Cardiac arrest/arrhythmia Cerebrovascular disease Peripheral vascular disease AIDS Cancer Body mass index (kg/m²) Pre-ESRD care Erythropoietin use before ESRD initiation Functional status at ESRD onset Inability to ambulate Inability to transfer Laboratory indices Hematocrit (%) Serum albumin (g/dl) Serum creatinine (mg/dl) GFR (ml/min per 1.73 m2)d Peritoneal dialysis 90 d after ESRD initiation

Non-Hispanica (n ⫽ 90,225)

Hispanica (n ⫽ 10,393)

Pb

61.8 (15.3)

59.2 (15.0)

⬍0.0001

53 47

52.8 47.2

0.71 0.71

60.7 34.1 5.2 42.3

87.1 4.6 8.3 61.3

⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001

48.7 73.2 6.7 7.3 26.5 9.3 33 6.6 9.2 14.8 0.6 5.2 25.6 (5.8)

65.3 67.5 2.4 2.8 20.4 5.6 29.7 3.7 6.6 13.2 0.2 2.4 25.8 (5.5)

⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 0.01

23.8

21.7

⬍0.0001

4.3 1.5

4.1 1.1

28.1 (5.3) 3.2 (0.7) 8.6 (3.8) 7.1 (2.8) 13.3

27.6 (5.3) 3.1 (0.7) 8.6 (3.9) 6.8 (2.7) 10.4

0.18 ⬍0.01 ⬍0.0001 ⬍0.0001 0.63 ⬍0.0001 ⬍0.0001

a

Source: Medical Evidence form 2728 for ESRD patients who started renal replacement therapy between May 1, 1995, and July 31, 1997. All values reported as mean ⫾ SD or % present. b P value by ␹2 test for categorical variables and t test for continuous variables. c Includes history of coronary artery disease, myocardial infarction, coronary artery bypass surgery, angioplasty, or abnormal angiography. d Estimated from the Modification of Diet in Renal Disease (MDRD) Study equation.

incomplete data were nearly identical to those of the study population. Table 2 provides a more detailed comparison of patients’ characteristics among Hispanic and racial subgroups. The largest subgroups were non-Hispanic whites, contributing to 60.7% of nonHispanics, and Hispanic whites, contributing to 87.1% of Hispanics. Among Hispanics, the prevalence of comorbid conditions was significantly greater among Hispanic whites and Hispanic blacks compared with Hispanic others. Serum creatinine levels and estimated GFR values were significantly lower among

Hispanic whites and Hispanic others compared with Hispanic blacks. Among non-Hispanics, the pattern of differences was similar to those among Hispanic categories with some exceptions. First, non-Hispanic whites were older but had lower prevalence of hypertension and diabetes compared with the other non-Hispanic categories. Second, although serum creatinine values before first dialysis were lowest for non-Hispanic whites, estimated GFR levels were significantly higher compared with non-Hispanic blacks and non-Hispanic others. Finally, non-Hispanic whites had the highest rates of peritoneal dialysis utilization.

J Am Soc Nephrol 16: 782–790, 2005


785

Table 2. Differences in baseline variables among ethnic and racial groups with ESRD at dialysis initiationa Variable

Hispanic Hispanic Hispanic Non-Hispanic Non-Hispanic Non-Hispanic White Black Other White Black Other (n ⫽ 9064) (n ⫽ 464) (n ⫽ 865) (n ⫽ 54,802) (n ⫽ 30,655) (n ⫽ 4768)

Demographic factors Age at ESRD onset (yr) 59.4 (14.9)e 56.4 (15.7) 59.4 (14.8) Gender Male 53e 56g 50 Female 47e 44g 50 Diabetes (cause of ESRD) 63e 48h 49i Comorbid conditions Diabetes (cause of ESRD or comorbidity) 67e 54h 49i Hypertension 72e 69g 15i e g Tobacco use (current or past) 2 3 1j Chronic obstructive lung disease 3e 3 0.6i c e h Coronary artery disease 22 20 6i Myocardial infarction 6e 6 2i Congestive heart failure 32e 29 8i Cardiac arrest or dysrhythmia 4e 4 1 Pericarditis 1 1 0.1i Cerebrovascular disease 7e 6 2i Peripheral vascular disease 14e 13h 4i AIDS 0.2f 1.6 0.3 Cancer 3e 3 0.1i Body mass index (kg/m2) 25.8 (5.5)e 25.3 (5.4)h 26.0 (6.3)i Erythropoietin use before ESRD onset 20.1e 20.4 18.0i Functional status at ESRD onset Inability to ambulate 4 5 0.6i Inability to transfer 1 1 0 Laboratory indices Hematocrit (%) 27.7 (5.2)e 27.3 (5.7) 27.3 (5.3) Serum albumin (g/dl) 3.1 (0.7)e 3.2 (0.7) 3.2 (0.7)j Serum creatinine (mg/dl) 8.5 (3.9)e 9.7 (4.5) 9.0 (4.0) 2 d GFR (ml/min per 1.73 m ) 6.8 (2.7)e 7.3 (3.2)h 6.4 (2.6) Dialysis modality at day 90 Peritoneal dialysis 10e 10 15

64.3 (14.7) 57 44 41

57.5 (15.5) 47 53 43

60.2 (14.8) 49 51 55

Pb

⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001

47 71 7 10 33 12 36 8 1 10 18 0.1 6 25.4 (5.6) 26.1

49 77 7 4 16 5 28 4 1 8 10 1.6 3 26.2 (6.2) 19.2

59 75 3 3 20 7 29 4 1 7 10 0.1 3 24.2 (5.2) 27.7

⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 0.0021 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001

4 1

4 2

3 1

⬍0.0001 ⬍0.0001

28.8 (5.2) 3.3 (0.6) 7.9 (3.1) 7.2 (2.7)

27.0 (5.5) 3.2 (0.7) 9.9 (4.5) 6.9 (2.9)

27.6 (5.5) 3.1 (0.7) 9.0 (3.9) 6.3 (2.6)

⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001

16

9

13

⬍0.0001

a

Source: Medical Evidence form 2728 for ESRD patients who started renal replacement therapy between May 1, 1995, and July 31, 1997. All values reported as mean ⫾ SD or % present. b P value by ␹2 test for categorical variables and ANOVA for continuous variables. c Includes history of coronary artery disease, myocardial infarction, coronary artery bypass surgery, angioplasty, or abnormal angiography. d Estimated from the MDRD Study equation. Comparison of Hispanic white versus non-Hispanic white, eP ⬍ 0.0001, fP ⬍ 0.01. Comparison of Hispanic black versus non-Hispanic black, gP ⬍ 0.0001, hP ⬍ 0.01. Comparison of Hispanic other versus non-Hispanic other, iP ⬍ 0.0001, jP ⬍ 0.01.

Likelihood of Having Selected Comorbid Conditions at ESRD Initiation among Ethnic Groups Table 3 illustrates the adjusted OR for selected comorbid conditions and markers of pre-ESRD care among Hispanic and non-Hispanic racial subgroups at ESRD initiation. Compared with non-Hispanic whites, the likelihood of diabetes was significantly greater for the remaining non-Hispanic and Hispanic subgroups. Hispanic whites had the greatest odds of having diabetes (OR, 2.43; 95% CI, 2.32 to 2.55), and Hispanic blacks (OR, 1.47; 95% CI, 1.22 to 1.77) and Hispanic others (OR, 1.40; 95% CI, 1.22 to 1.60) also had significantly higher odds. Contrasting, the likelihood of coronary artery disease was significantly lower for all Hispanic groups compared with non-Hispanic whites (Hispanic white: OR, 0.65; 95% CI, 0.62 to 0.69; Hispanic black: OR, 0.71; 95% CI, 0.56 to 0.91; and Hispanic

other: OR, 0.27; 95% CI, 0.20 to 0.36). Moreover, non-Hispanic black and non-Hispanic other categories had similar likelihood of coronary artery disease to that of Hispanic subgroups. As anemia is an important predictor of mortality in ESRD patients, we examined differences in prevalence of anemia at ESRD onset among race/ethnic subgroups. Compared with non-Hispanic whites, the likelihood of severe anemia (defined as hematocrit ⱕ28%) was significantly greater for all Hispanic groups (Hispanic whites: OR, 1.28; 95% CI, 1.22 to 1.34; Hispanic blacks: OR, 1.44; 95% CI, 1.20 to 1.73; Hispanic others: OR, 1.34; 95% CI, 1.17 to 1.54). Interestingly, non-Hispanic blacks and non-Hispanic other groups also had an increased likelihood of being anemic at ESRD onset (non-Hispanic blacks: OR, 1.52; 95% CI, 1.47 to 1.56; non-Hispanic others: OR, 1.27; 95% CI, 1.19 to 1.34). The prescribing of erythropoietin before ESRD

786


J Am Soc Nephrol 16: 782–790, 2005

Table 3. Likelihood of having selected comorbid conditions or receiving treatment with erythropoietin at ESRD onset among ethnic groups Coronary Diseasea

Diabetesa Ethnic and Racial Subgroup

Non-Hispanic white (n ⫽ 54,802) Non-Hispanic black (n ⫽ 30,655) Non-Hispanic other (n ⫽ 4768) Hispanic white (n ⫽ 9064) Hispanic black (n ⫽ 464) Hispanic other (n ⫽ 865)

Hematocrit ⱕ28%a

Pre-ESRD Erythropoietin Usea

Serum Albumin ⱕ3.5 g/dla

OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

1.00b 1.22 1.68 2.43 1.47 1.40

1.18-1.26 1.58-1.79 2.32-2.55 1.22-1.77 1.22-1.60

1.00b 0.53 0.66 0.65 0.71 0.27

0.51–0.55 0.61–0.72 0.62–0.69 0.56–0.91 0.20–0.36

1.00b 1.52 1.27 1.28 1.44 1.34

1.47–1.56 1.19–1.34 1.22–1.34 1.20–1.73 1.17–1.54

1.00b 0.71 1.10 0.82 0.77 0.69

0.68–0.73 1.03–1.18 0.78–0.86 0.62–0.96 0.58–0.82

1.00b 1.16 1.26 1.22 1.02 1.00

1.12–1.20 1.18–1.35 1.16–1.28 0.84–1.24 0.87–1.16

a Logistic regression model, adjusted for age at study start, gender, hypertension, cardiac arrest/arrhythmia, peripheral vascular and cerebrovascular disease, congestive heart failure, tobacco use, chronic lung disease, AIDS, neoplasm, alcohol dependence, body mass index, estimated GFR, functional status (inability to walk and transfer), diabetes, coronary artery disease, hematocrit, pre-ESRD erythropoietin use, and serum albumin. OR, odds ratio; CI, confidence interval. b Reference group.

initiation was also captured from the CMS Form and allowed comparisons of pre-ESRD care delivery. Compared with nonHispanic whites, the use of erythropoietin was significantly lower among all Hispanic groups (Hispanic whites: OR, 0.82; 95% CI, 0.78 to 0.86; Hispanic blacks: OR, 0.77; 95% CI, 0.62 to 0.96; and Hispanic other: OR, 0.69; 95% CI, 0.58 to 0.82). The likelihood of erythropoietin use was equally low for non-Hispanic blacks (OR, 0.71; 95% CI, 0.68 to 0.73) but significantly higher for non-Hispanic others (OR, 1.10; 95% CI, 1.03 to 1.18). Finally, we compared the likelihood of having a low serum albumin at ESRD onset among groups. Compared with nonHispanic whites, only Hispanic whites were more likely to be hypoalbuminemic at ESRD onset (OR, 1.22; 95% CI, 1.16 to 1.28) within the Hispanic groups. However, non-Hispanic blacks and non-Hispanic others had significantly greater likelihood of hypoalbuminemia (non-Hispanic blacks: OR, 1.16; 95% CI, 1.12 to 1.20; non-Hispanic others: OR, 1.26; 95% CI, 1.18 to 1.35) compared with non-Hispanic whites.

Survival Differences between Hispanic and Non-Hispanic Patients during Follow-Up Of the 100,618 incident patients, 25% patients died during the follow-up period. Adjusted Cox survival curves were estimated for non-Hispanic and Hispanic patients and are shown in Figures 1 and 2. Overall survival was significantly poorer for non-Hispanics compared with Hispanics in both diabetic (Figure 1) and nondiabetic (Figure 2) cohorts. For diabetic patients, the survival curves begin to diverge early on and continue to diverge thereafter, with significantly poorer survival for nonHispanic compared with Hispanic patients. For those without diabetes, the pattern was similar, although less impressive, with divergence of the survival curves occurring after 6 mo of follow-up.

Mortality Differences between Hispanic and Non-Hispanic Patients The overall death rates and unadjusted and adjusted hazard ratios with 95% CI for Hispanic and non-Hispanic patients are

Figure 1. Adjusted Cox survival curves for new diabetic ESRD patients by Hispanic ethnicity. All models adjusted for age at study start, gender, race, hypertension, coronary artery disease, cardiac arrest/arrhythmia, peripheral vascular and cerebrovascular disease, congestive heart failure, tobacco use, chronic lung disease, AIDS, neoplasm, alcohol dependence, body mass index, serum albumin, hematocrit, estimated GFR, functional status (inability to walk and transfer), and pre-ESRD erythropoietin use.

given in Table 4. Unadjusted death rates were significantly lower for Hispanics (19.2 and 14.7 deaths per 100 person-years for those with and without diabetes, respectively) compared with non-Hispanics (26 and 22.7 per 100 person-years at risk). For those with diabetes, the unadjusted relative risk (RR) of death for Hispanics compared with non-Hispanics was 0.74 (95% CI, 0.70 to 0.78). With adjustment for case mix, RR of death was virtually identical at 0.70 (95% CI, 0.66 to 0.74). Similarly, for those without diabetes, the unadjusted and adjusted RR of death for Hispanics was 0.65 (95% CI, 0.60 to 0.71) and 0.83 (95% CI, 0.77 to 0.91), respectively. To characterize these relationships more fully, we evaluated

J Am Soc Nephrol 16: 782–790, 2005

Figure 2. Adjusted Cox survival curves for new nondiabetic ESRD patients by Hispanic ethnicity. All models adjusted for age at study start, gender, race, hypertension, coronary artery disease, cardiac arrest/arrhythmia, peripheral vascular and cerebrovascular disease, congestive heart failure, tobacco use, chronic lung disease, AIDS, neoplasm, alcohol dependence, body mass index, serum albumin, hematocrit, estimated GFR, functional status (inability to walk and transfer), and pre-ESRD erythropoietin use.

the mortality risks for each ethnic-racial subclassification (Table 5). For those with diabetes, the unadjusted RR of death was significantly lower for all Hispanic subgroups compared with non-Hispanic whites: Hispanic whites (RR ⫽ 0.60; 95% CI, 0.56 to 0.64), Hispanic blacks (RR ⫽ 0.61; 95% CI, 0.47 to 0.80), and Hispanic others (RR ⫽ 0.79; 95% CI, 0.67 to 0.94). It is interesting that these RR estimates were similar in magnitude to those of non-Hispanic blacks (RR ⫽ 0.60; 95% CI, 0.57 to 0.62) and non-Hispanic others (RR ⫽ 0.64; 95% CI, 0.59 to 0.69). With adjustment, the mortality advantage remained significant for all groups except for the Hispanic other category. For those without diabetes, the mortality patterns were similar, with Hispanic whites (RR ⫽ 0.76; 95% CI, 0.69 to 0.83), non-Hispanic blacks (RR ⫽ 0.91; 95% CI, 0.87 to 0.96), and non-Hispanic others (RR ⫽ 0.62; 95% CI, 0.54 to 0.70) experiencing significantly lower mortality risks compared with nonHispanic whites. However, in this group, the Hispanic black and Hispanic other categories did not experience a survival advantage (RR ⫽ 0.89; 95% CI, 0.65 to 1.22; and RR ⫽ 1.01; 95% CI, 0.84 to 1.22, respectively).

Discussion In this study, we describe the mortality risk patterns of an incident dialysis cohort in the United States with contemporaneous consideration of ethnicity and race. Overall, mortality risks for Hispanic patients were significantly lower than that of non-Hispanic patients, although this pattern was not consistent across all subgroups. Among those with diabetes, the survival advantage of Hispanic white and black patients over nonHispanic whites did not extend to the Hispanic other category


787

after adjustment. Among those without diabetes, neither patients who were categorized as Hispanic black or Hispanic other achieved a survival advantage over non-Hispanic whites. This study highlights significant variation in mortality risks among Hispanic subgroups initiating dialysis therapy in the United States, a finding that was not explained by adjustment for differences in case mix. Moreover, it demonstrates an important effect modification of race on the ethnicity–survival relationship, underscoring the need for simultaneous inclusion of these indicators in future analyses when evaluating clinical outcomes. The overall survival advantage of Hispanics over non-Hispanics in this study might well be attributed to lower comorbidity burden at ESRD onset. The prognostic impact of each of these comorbid conditions has been demonstrated consistently in previous mortality analyses (14,15). When these differences were accounted for in the adjusted analysis, the lower mortality risk for Hispanics remained. A novel and more perplexing observation from this study is the significant variation in mortality rates among Hispanic subgroups. Hispanic whites experienced the lowest mortality rates among all Hispanic groups, a finding that was present in those both with and without diabetes. Lower mortality risks were also observed for Hispanic blacks, but these were seen only among those with diabetes. Finally, no mortality advantage was seen for Hispanic others, who had similar risk for death to those of non-Hispanic whites. This was surprising as cardiovascular comorbidity levels were substantially lower and BMI levels were significantly higher in the Hispanic other category compared with the remaining Hispanic groups—findings that favored increased survival (16 – 18). Although differences in case mix may have contributed to these findings, it should also be noted that the lack of survival advantage among the Hispanic other category may reflect the heterogeneity of this group encompassing people with diverse genetic backgrounds. Differences in access to and delivery of health care between Hispanic and non-Hispanic cohorts as well as among Hispanic subgroups before ESRD start could also have contributed to these mortality differences. Several major indicators of preESRD care, including hematocrit level at ESRD initiation, serum albumin concentration, and prescribed erythropoietin use before initiation of dialysis, were significantly lower among Hispanics compared with non-Hispanics. These findings concur with the observations of Kausz et al. (5), who, in a similar study, found that Hispanics were 47% more likely to be initiated late on dialysis (GFR ⬍5 ml/min at initiation of dialysis) compared with whites. Moreover, many of these components of pre-ESRD care have strong prognostic significance in survival analysis (19 –22). Despite adjusting for several pre-ESRD care indicators, the overall mortality advantage of Hispanics over non-Hispanics remained. Similarly, in the subgroup comparisons, the higher mortality rates among the Hispanic other category persisted, having taken into consideration baseline differences in pre-ESRD care among groups. It might be argued that the higher relative mortality risks among the Hispanic other group might reflect unequal delivery of pre-ESRD care. The low rates of pre-ESRD erythropoietin use and the low levels of residual

788


J Am Soc Nephrol 16: 782–790, 2005

Table 4. Mortality risks of Hispanic and non-Hispanic ESRD patients who were initiated on dialysis in the United Statesa Diabetesb

Ethnic Group

Hazard No. of % Deaths per Ratioc Patients Deaths 100 Person-Years Unadjusted

No Non-Hispanic (n ⫽ 49,867) Hispanic Yes Non-Hispanic (n ⫽ 50,751) Hispanic

46,273 3,594 43,952 6,799

23.9 16.9 27 21.3

22.7 14.7 26 19.2

1.00 0.65 1.00 0.74

95% CI

Pc

Reference 0.60–0.71 ⬍0.0001 Reference 0.70–0.78 ⬍0.0001

Hazard Ratioc Adjustedd 1.00 0.83 1.00 0.70

Pc

95% CI

Reference 0.77–0.91 ⬍0.0001 Reference 0.66–0.74 ⬍0.0001

a

Source: Medical Evidence form 2728 and Death Notification form 2746 for patients who were initiated on dialysis between May 1, 1995, and July 31, 1997. b Diabetes as a cause of ESRD or as a comorbid condition. c Hazard ratio and P value by Cox proportional hazards model. d Model adjusted for age at study start, gender, race, hypertension, coronary artery disease, cardiac arrest/arrhythmia, peripheral vascular and cerebrovascular disease, congestive heart failure, tobacco use, chronic lung disease, AIDS, neoplasm, alcohol dependence, body mass index, serum albumin, hematocrit, estimated GFR, functional status (inability to walk and transfer), and pre-ESRD erythropoietin use.

Table 5. Mortality risks of ethnic and racial subgroups who were initiated on dialysis in the United Statesa Diabetesb

Ethnic Group

No Non-Hispanic white (n ⫽ 49,867) Non-Hispanic black Non-Hispanic other Hispanic white Hispanic black Hispanic other Yes Non-Hispanic white (n ⫽ 50,751) Non-Hispanic black Non-Hispanic other Hispanic white Hispanic black Hispanic other

No. Deaths Hazard % of per 100 Ratioc Deaths Patients Person-Years (Unadjusted) 28,848 15,497 1,928 2,938 211 445 25,954 15,158 2,840 6,126 253 420

27.3 18.9 13.1 15.4 18 26.5 31 21 22.2 20.6 21.7 31.4

27.1 16.8 11.4 13.6 16.9 19.7 31.3 18.7 19.9 18.7 19.1 24.8

1.00 0.63 0.43 0.51 0.61 0.75 1.00 0.60 0.64 0.60 0.61 0.79

95% CI

Pc

Hazard Ratioc,d (Adjusted)

95% CI

Pc

Reference 0.60–0.65 0.38–0.48 0.46–0.56 0.45–0.84 0.62–0.90 Reference 0.57–0.62 0.59–0.69 0.56–0.64 0.47–0.80 0.67–0.94

Reference ⬍0.0001 ⬍0.0001 ⬍0.0001 0.0025 0.0016 Reference ⬍0.0001 ⬍0.0001 ⬍0.0001 0.0003 0.0086

1.00 0.91 0.62 0.76 0.89 1.01 1.00 0.68 0.67 0.65 0.67 0.96

Reference 0.87–0.96 0.54–0.70 0.69–0.83 0.65–1.22 0.84–1.22 Reference 0.65–0.71 0.61–0.72 0.61–0.69 0.52–0.88 0.81–1.15

Reference 0.0001 ⬍0.0001 ⬍0.0001 0.4655 0.9034 Reference ⬍0.0001 ⬍0.0001 ⬍0.0001 0.0034 0.6820

a

Source: Medical Evidence form 2728 and Death Notification form 2746 for ESRD patients who started renal replacement therapy between May 1, 1995, and July 31, 1997; period of follow-up, 2 yr. b Diabetes as a cause of ESRD or as a comorbidity. c Hazard ratio and P value by Cox proportional hazards model. d Model adjusted for age at study start, gender, race, hypertension, coronary artery disease, cardiac arrest/arrhythmia, peripheral vascular and cerebrovascular disease, congestive heart failure, tobacco use, chronic lung disease, AIDS, neoplasm, alcohol dependence, body mass index, serum albumin, hematocrit, estimated GFR, functional status (inability to walk and transfer), and pre-ESRD erythropoietin use.

renal function at dialysis initiation among patients in the Hispanic other category are certainly suggestive of this. Nevertheless, the adjusted analysis also confirmed higher-than-expected death rates in the Hispanic other category, suggesting that factors other than those measured contributed to these differences. Although several studies have compared the mortality risks of Hispanics and non-Hispanics in the general population, few have compared survival differences in ESRD cohorts (23–25). Those published have been limited in generalizability as a result of center effect or selection bias (23,24). To our knowledge, only one other study has explored the mortality risks of Hispanic patients who undergo dialysis therapy. Frankenfield et al. (4), in an analysis of data from the ESRD Clinical Performance Measures Project, found significantly lower mortality among Hispanic compared with non-Hispanic patients who

were treated with HD after 1 yr of follow-up. Although this study had several strengths—large sample size and nationally representative—it also had a number of limitations that could potentially confound mortality analysis that is unique to this population. First, their analyses were based on a cross-sectional sample of “prevalent” dialysis patients. The survival bias of prevalent ESRD cohorts is a well-described limitation in mortality analyses of this nature. Second, analyses were restricted to HD patients, thereby reducing its applicability to those who were treated with peritoneal dialysis. Furthermore, the availability of clinical and laboratory data on comorbid conditions recorded at ESRD onset and surrogate markers of pre-ESRD care were lacking. Our study, in contrast, overcame many of these deficits. The inclusion of an incident dialysis population that is nationally representative; the adjustment for a broad list of sociodemographic, comorbid, and pre-ESRD care indicators;

J Am Soc Nephrol 16: 782–790, 2005

and finally the extension of the mortality comparisons to Hispanic subgroups are signal strengths of our approach. A “survival advantage” for Hispanics has been observed in the general population, as well as in relation to specific disease (25–28). This advantage is observed despite a lower socioeconomic status, higher poverty rates, less education, and less health insurance, all factors that are potentially associated with higher mortality and morbidity rates. The reasons for this so called “Hispanic paradox” are poorly understood, but several hypotheses have been postulated. They include greater family support systems, more favorable health behaviors, genetic factors, and the “salmon-bias” theory, which postulates lower mortality rates among Hispanics in the United States on the basis that older Hispanics return to their native birthplace to die and are as a result not included in the US mortality statistics (29). Several possible explanations could account for the overall lower mortality rates among Hispanics despite their adverse risk profile documented at ESRD initiation. First, factors that are associated with a Hispanic survival benefit in the general population may be operating in patients who are treated with ESRD. Second, reduced access to transplantation for Hispanics and lower rates of renal transplantation for those who are wait-listed could explain the “apparent” survival advantage of Hispanics over non-Hispanics who remain in the dialysis pool (30 –35). Transplantation was a censoring event in the current analysis ensuring that any observed survival advantage was attributed only to characteristics that occurred before or during the dialysis period. Finally, the contribution of genetic polymorphisms toward a Hispanic survival advantage cannot be ignored. A lower prevalence of complex trait diseases such as coronary artery disease as a result of reduced gene expression may exert considerable protection in Hispanic populations (36). This study is not without limitations. First, the possibility of misclassification of ethnicity status should not be ignored. The designation of Hispanic ethnicity is typically carried out by personnel at the dialysis unit. We submit that assignment of ethnic status is not based on self-designation or through a systematic assessment of a patient’s background and may result in misclassification. However, the assignment of ethnicity is usually performed by members of the dialysis team who have interacted with the patient and their caregivers on several health-related, social, and financial issues, for which appropriate classification of ethnic status may be required. As such, we believe that a serious misclassification is unlikely, and, if it occurred, it is likely to be random in nature. Second, differential underreporting of comorbid conditions on the CMS 2728 form may bias mortality comparisons among groups and lead to biased estimates (37). For this reason, we repeated the regression models adjusting only for objective measures of comorbid status, including serum albumin, hematocrit, serum creatinine at ESRD initiation (an indirect measure of nutrition), BMI, and residual renal function, and found similar results. Third, the CMS document, while capturing the presence of comorbid disease, does not capture disease severity, which may have differed between Hispanics and non-Hispanics as well as among Hispanic subgroups. Although this is a potential concern, it should be noted that recent comparisons of different comorbidity instruments have


789

demonstrated no incremental benefit of those that use disease severity -grading systems over those that do not (38). In conclusion, the current study demonstrates that the overall survival of new ESRD patients of Hispanic ethnicity is substantially greater than that of non-Hispanics, with a 17% lower adjusted mortality risk among those without diabetes and a 30% lower adjusted mortality risk among those with diabetes. However, this survival advantage is not consistent across all Hispanic subgroups, with Hispanic whites and Hispanic blacks experiencing the lowest and Hispanic others experiencing the highest mortality rates. Although the mortality risks for Hispanic whites and Hispanic blacks were significantly lower than those of non-Hispanic Whites, they were almost equal to those of non-Hispanic blacks and other non-Hispanic groups. These differences in mortality outcomes cannot be explained easily by differences in baseline comorbidity profiles among groups or differences in transplantation rates during follow-up. Consequently, greater efforts should be invested in evaluating the contribution of access to dialysis therapy and, once on dialysis, access to transplant waiting list as possible causes for these differences. Moreover, the contribution of genetic polymorphisms to these disparate outcomes requires further exploration. Finally, given the strong effect modification of race on the Hispanic–survival relationship, we suggest that both race and ethnicity be considered as contemporaneous predictor variables in future mortality analyses. The disparity in mortality outcomes across Hispanic subgroups as demonstrated in this study should urge health care providers and policy makers to improve accessibility and delivery of medical care to all minority groups with chronic kidney disease.

Acknowledgments A.G.S. is supported by a National Scientist Development Grant (0335317N) from the American Heart Association. A portion of this work was presented in part at the annual meeting of the American Society of Nephrology, San Diego, California, November 2003, and has been published in abstract form (J Am Soc Nephrol 14: 715A, 2003).

References 1. Sundquist J, Winkleby MA, Pudaric S: Cardiovascular disease risk factors among older black, Mexican-American, and white women and men: An analysis of NHANES III, 1988 –1994. Third National Health and Nutrition Examination Survey. J Am Geriatr Soc 49: 109 –116, 2001 2. US Census Bureau: Difference in Population by Race and Hispanic or Latino Origin, for the United States: 1990 to 2000, Washington, DC, US Census Bureau, 2002 3. United States Renal Data System: USRDS Annual Data Report, 2002, Bethesda, National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases, 2002 4. Frankenfield D, Rocco M, Roman S, McClellan W: Survival advantage for adult Hispanic hemodialysis patients? Findings from the End-Stage Renal Disease Clinical Performance Measures Project. J Am Soc Nephrol 14: 180 –186, 2003 5. Kausz AT, Obrador GT, Arora P, Ruthazer R, Levey AS, Pereira BJ: Late initiation of dialysis among women and

790

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.


ethnic minorities in the United States. J Am Soc Nephrol 11: 2351–2357, 2000 United States Renal Data System. Patient characteristics at the start of ESRD: Data from the HCFA Medical Evidence Form. Am J Kidney Dis 34:S63–S73, 1999 Roys EC, Port FK, Agodoa LYC, Meyers-Purkiss A, Brown PL, Jones CA, Daugirdas JT, Pereira BJG, Golper TA, Wolfe RA: Validation of medical evidence form with DMMS data showing relative importance as predictor of mortality [Abstract]. J Am Soc Nephrol 10: 255A, 1999 Obrador GT, Roberts T, St Peter WL, Frazier E, Pereira BJ, Collins AJ: Trends in anemia at initiation of dialysis in the United States. Kidney Int 60: 1875–1884, 2001 Xue JL, Chen SC, Ebben JP, Constantini EG, Everson SE, Frazier ET, Agodoa LY, Collins AJ: Peritoneal and hemodialysis: I. Differences in patient characteristics at initiation. Kidney Int 61: 734 –740, 2002 Xue JL, Everson SE, Constantini EG, Ebben JP, Chen SC, Agodoa LY, Collins AJ: Peritoneal and hemodialysis: II. Mortality risk associated with initial patient characteristics. Kidney Int 61: 741–746, 2002 Owen WF Jr: Racial differences in incidence, outcome, and quality of life for African-Americans on hemodialysis. Blood Purif 14: 278 –285, 1996 Pei YP, Greenwood CM, Chery AL, Wu GG: Racial differences in survival of patients on dialysis. Kidney Int 58: 1293–1299, 2000 Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130: 461– 470, 1999 United States Renal Data System: Comorbid conditions and correlations with mortality risk among 3,399 incident hemodialysis patients. Am J Kidney Dis 20:32–38, 1992 Ganesh SK, Hulbert-Shearon T, Port FK, Eagle K, Stack AG: Mortality differences by dialysis modality among incident ESRD patients with and without coronary artery disease. J Am Soc Nephrol 14: 415– 424, 2003 Kopple JD, Zhu X, Lew NL, Lowrie EG: Body weight-forheight relationships predict mortality in maintenance hemodialysis patients. Kidney Int 56: 1136 –1148, 1999 Port FK, Ashby VB, Dhingra RK, Roys EC, Wolfe RA: Dialysis dose and body mass index are strongly associated with survival in hemodialysis patients. J Am Soc Nephrol 13: 1061–1066, 2002 Glanton CW, Hypolite IO, Hshieh PB, Agodoa LY, Yuan CM, Abbott KC: Factors associated with improved short term survival in obese end stage renal disease patients. Ann Epidemiol 13: 136 –143, 2003 Xue JL, St Peter WL, Ebben JP, Everson SE, Collins AJ: Anemia treatment in the pre-ESRD period and associated mortality in elderly patients. Am J Kidney Dis 40: 1153– 1161, 2002 Fink J, Blahut S, Reddy M, Light P: Use of erythropoietin before the initiation of dialysis and its impact on mortality. Am J Kidney Dis 37: 348 –355, 2001 Ma JZ, Ebben J, Xia H, Collins AJ: Hematocrit level and associated mortality in hemodialysis patients. J Am Soc Nephrol 10: 610 – 619, 1999 Stack AG: Impact of timing of nephrology referral and

J Am Soc Nephrol 16: 782–790, 2005

23.

24.

25. 26. 27.

28.

29.

30.

31.

32. 33.

34.

35.

36.

37.

38.

pre-ESRD care on mortality risk among new ESRD patients in the United States. Am J Kidney Dis 41: 310 –318, 2003 Pugh JA, Tuley MR, Basu S: Survival among MexicanAmericans, non-Hispanic whites, and African-Americans with end-stage renal disease: The emergence of a minority pattern of increased incidence and prolonged survival. Am J Kidney Dis 23: 803– 807, 1994 Medina RA, Pugh JA, Monterrosa A, Cornell J: Minority advantage in diabetic end-stage renal disease survival on hemodialysis: Due to different proportions of diabetic type? Am J Kidney Dis 28: 226 –234, 1996 Sorlie PD, Backlund E, Johnson NJ, Rogot E: Mortality by Hispanic status in the United States. JAMA 270: 2464–2468, 1993 Staub L, Morgenstern LB: Stroke in Hispanic Americans. Neurol Clin 18: 291–307, 2000 Schiff M, Key CR, Gilliland FD, Becker TM: Ethnic differences in uterine corpus cancer incidence and mortality in New Mexico’s American Indians, Hispanics and non-Hispanic whites. Int J Epidemiol 26: 249 –255, 1997 Canto JG, Taylor HA Jr, Rogers WJ, Sanderson B, Hilbe J, Barron HV: Presenting characteristics, treatment patterns, and clinical outcomes of non-black minorities in the National Registry of Myocardial Infarction 2. Am J Cardiol 82: 1013–1018, 1998 Liao Y, Cooper RS, Cao G, Durazo-Arvizu R, Kaufman JS, Luke A, McGee DL: Mortality patterns among adult Hispanics: Findings from the NHIS, 1986 to 1990. Am J Public Health 88: 227–232, 1998 Wolfe RA, Ashby VB, Milford EL, Bloembergen WE, Agodoa LY, Held PJ, Port FK: Differences in access to cadaveric renal transplantation in the United States. Am J Kidney Dis 36: 1025–1033, 2000 Kasiske BL, London W, Ellison MD: Race and socioeconomic factors influencing early placement on the kidney transplant waiting list. J Am Soc Nephrol 9: 2142–2147, 1998 Gadegbeku C, Freeman M, Agodoa L. Racial disparities in renal replacement therapy. J Natl Med Assoc 94:45S–54S, 2002 Soucie JM, Neylan JF, McClellan W: Race and sex differences in the identification of candidates for renal transplantation. Am J Kidney Dis 19: 414 – 419, 1992 Gaylin DS, Held PJ, Port FK, Hunsicker LG, Wolfe RA, Kahan BD, Jones CA, Agodoa LY: The impact of comorbid and sociodemographic factors on access to renal transplantation. JAMA 269: 603– 608, 1993 Held PJ, Pauly MV, Bovbjerg RR, Newmann J, Salvatierra O Jr: Access to kidney transplantation. Has the United States eliminated income and racial differences? Arch Intern Med 148: 2594 –2600, 1988 Lee JH, Tang MX, Schupf N, Stern Y, Jacobs DM, Tycko B, Mayeux R: Mortality and apolipoprotein E in Hispanic, African-American, and Caucasian elders. Am J Med Genet 103: 121–127, 2001 Longenecker JC, Coresh J, Klag MJ, Levey AS, Martin AA, Fink NE, Powe NR: Validation of comorbid conditions on the end-stage renal disease medical evidence report: The CHOICE study. Choices for Healthy Outcomes in Caring for ESRD. J Am Soc Nephrol 11: 520 –529, 2000 Van Manen JG, Korevaar JC, Dekker FW, Boeschoten EW, Bossuyt PM, Krediet RT: How to adjust for comorbidity in survival studies in ESRD patients: A comparison of different indices. Am J Kidney Dis 40: 82– 89, 2002