Racial Differences in Survival of Incident Home Hemodialysis and Kidney Transplant Patients

Abstract

Background

Previous studies have indicated that patients on maintenance hemodialysis(HD) have worse survival compared to kidney transplant(KTx) recipients. However, none of these studies have compared mortality of the United States(US) patients using alternative dialysis modalities such as home HD with KTx recipients.

Methods

Comparing patients who started home HD with those who received kidney transplantation in the US between 2007–2011, we created a 1:1 propensity-matched(PS) cohort of 4,000 patients and examined the association between treatment modality and all-cause mortality using Cox proportional hazard models.

Results

The mean±SD age of the PS-matched home HD and KTx patients at baseline were 54±15 years and 54±14 years, 65% were male(both groups), 70% and 72% of patients were whites and 19% were African American(both groups), respectively. Over 5-years of follow-up, home HD patients had 4-times higher mortality risk compared to KTx recipients in the entire patient population (hazard ratio(HR):4.06,95%confidence interval(CI):3.27–5.04);total event number=411), and similar difference was found across each race stratum. However, during the first year of therapy, while the white home HD patients had higher mortality risk (HR:4.21,95%CI:3.10–5.73;total event number=332) compared to their KTx counterparts, there was no significant difference in mortality risk between African American home HD and KTx patients (HR:1.62,95%CI:0.77–3.39;total event number=55). This result was consistent across different types of kidney donors.

Conclusions

Home HD patients appear to have 4 times higher mortality compared to KTx recipients regardless of the type of kidney donor. Further studies are needed to understand the reasons underlying racial differenes during the first year of therapy.

Introduction

Kidney transplantation (KTx) is the treatment of choice for patients with end stage renal disease (ESRD). Several studies have compared survival of waitlisted dialysis patients with KTx recipients.^1–8 One of the largest study of 230,000 dialysis patients showed that mortality was significantly lower among patients who received a KTx compared with transplant wait-listed dialysis patients (3.8 vs. 6.3/100 patient-years).⁸

Previous studies showed that home hemodialysis (home HD) provides better survival than in-center HD. Marshall et al⁹ examined 865 home HD patients and 21,184 in-center HD patients over 72,052 patient-years and found that home HD was associated with 47% lower mortality risk, and Weinhandl et al¹⁰ reported 13% lower mortality risk in 1,873 home HD patients compared with 9,365 matched (1:5) patients. In addition, in a recent 1:10 propensity score (PS) matched analysis of patients from the United States, France and Canada, home HD patients had a 45% lower mortality risk compared to in-center HD patients.¹¹

It is currently unknown whether KTx provides better survival than home HD in ESRD patients. To the best of our knowledge, only two Canadian studies compared the survival of home HD patients with KTx recipients.^12–14 Pauly et al¹² used data from two regional programs in Canada to compare survival between patients treated with nocturnal home HD and patients who received a KTx from either a deceased or a living donor as reported in the US Renal Data System. Nocturnal home HD patients were randomly matched to patients who received either of the 2 transplant modalities in a 1:3:3 ratio; 177 nocturnal home HD patients were matched to 531 deceased and 531 living transplant recipients and followed for up to 12.4 years. In adjusted models, there was no difference in survival between nocturnal home HD patients and deceased transplant recipients (HR=0.87, 95% CI: 0.50–1.51); however, patients who received a living donor transplant had a 41% better survival compared to nocturnal home HD patients (HR=0.51, 95% CI: 0.28–0.91).¹² Furthermore, in another study which compared 173 Canadian nocturnal home HD patients to 1,517 Canadian KTx recipients from the same institution, patients with a KTx had a reduced risk of treatment failure and death compared with home HD patients.¹⁴

There are several reasons why the comparative effectiveness of the two modalities may differ in the United States; to our knowledge there are no such studies comparing home HD with KTx in the United States. There is a high prevalence of low-flow systems such as NxStage in the United States, which provide lower solute clearances than conventional HD machines used in Canada. In addition, none of the platforms in the United States have been approved for nocturnal HD, and hence most home HD patients undergo short, daily dialysis. Finally, the risk of death in patients undergoing HD or with KTx is in general higher in the United States than in Canada.^15,16

We examined the survival of incident patients in a large, nationally representative contemporary cohort of patients from the United States to test the hypothesis that patients undergoing home HD have a higher risk of mortality compared to those receiving a KTx.

Materials and Methods

Data Source and Cohort Definition

The study cohort comprised incident home HD patients treated by one of the largest dialysis providers in the United States, and incident KTx recipients transplanted between January 1^st 2007 and December 31^st 2011 in the entire US. Pertinent data for the two groups were obtained from electronic medical records from the large dialysis provider and from the United States Renal Data System, respectively. The study was approved by the Institutional Review Committees of the Los Angeles Biomedical Research Institute at Harbor-UCLA, University of California Irvine Medical Center, University of Washington and University of Tennessee Health Science Center.

Home Hemodialysis cohort

The original source population was a cohort of 208,820 incident (newly initiated) dialysis patients. Patients were included in the cohort if they were ≥ 18 years old. Patients were excluded if they did not receive dialysis treatment for at least 60 days or did not have any treatment with home HD over their duration of follow up. The detailed description of dialysis modality assignment is discussed elsewhere.¹⁷ Our final home HD cohort included 2,830 patients (Figure 1).

Figure 1.

Flow chart of patients’ selection

Kidney transplant cohort

The United States Renal Data System includes all patients who received a kidney transplant between 1963 and 2012. Patients ≥ 18 years old at time of transplant and who received their first kidney between January 1^st 2007 and December 31^st 2011 were included in our cohort. Patients were excluded if they were represented in the home HD cohort yielding a cohort with 73,976 KTx patients (Figure 1).

For the main analyses we created a PS (1:1) matched cohort consisting of 2,000 home HD and 2,000 KTx patients (Figure 1).

Exposure, Covariates, Outcome

Data on age, gender, race/ethnicity, ESRD etiology and vintage, access type, insurance, body mass index, serum albumin, blood hemoglobin and coexisting conditions was obtained from the two data sources and refined. The following nine coexisting conditions were considered: diabetes mellitus, hypertension, alcohol abuse, atherosclerotic heart disease, other cardiac disease (pericarditis and cardiac arrhythmia), congestive heart failure, cerebrovascular disease, chronic obstructive pulmonary disease, and malignancy. Exposure was defined as home HD versus KTx, and the outcome was all-cause mortality.

Statistical Analysis

Data were summarized using proportions, means ± SD, or median (interquartile range (IQR)) as appropriate. PS matching was used to account for baseline differences arising from dissimilarities in clinical and demographic characteristics of home HD and KTx patients. We created PS matched cohorts for all patients and in race strata (African-American and white). We additionally PS matched home HD patients to KTx recipients of specific KTx donor type [living, deceased, standard criterion donor (SCD), extended criterion donor (ECD), donor after cardiac death (DCD), non-DCD, or donor with Kidney Donor Profile Index¹⁸ (KDPI)>20 or ≤20]. STATA’s “psmatch2” command suite was used to generate the 1:1 PS-matched cohorts using nearest neighbor matching without replacement. The following variables were included in a logistic regression model to calculate the propensity scores: age, gender, race/ethnicity, primary insurance, type of vascular access at the time of transplantation/home HD, cause of ESRD, previous time on ESRD, body mass index, blood hemoglobin, serum albumin, comorbidities at baseline (diabetes, hypertension, cardiovascular disease, atherosclerotic heart disease, heart failure, cerebrovascular disease, chronic obstructive pulmonary disease, malignancy, and alcohol abuse). Home HD and KTx patients before and after matching were compared using standardized differences in all PS matched cohorts.¹⁹ Figure S1 shows the receiver operating characteristic curve of the calculated propensity score compared to actual modality in the entire cohort.

Associations between renal replacement modalities (home HD vs KTx) and all-cause mortality were assessed using the Kaplan-Meier method, and Cox proportional hazards models (for time to event analyses). Models in PS-matched cohorts were not additionally adjusted for covariates. For the main analyses, the start of the follow-up period was the start date of home HD modality or the date of kidney transplantation. Patients were followed until date of death, date of censoring [transfer to a different dialysis modality, kidney transplantation, transfer to a different facility or other reason for home HD patients, or date of allograft loss (re-transplantation, first date of dialysis) for KTx patients], or the end of the follow-up period (December 31^st, 2011) (Table S17). In sensitivity analyses, we used an alternative censoring method where home HD patients were not censored at time of transfer to a different dialysis modality and continued to be followed until death, end of follow-up, or for other causes of censoring. As a significant proportion of home HD patients were transplanted during the follow-up period, there is a significant degree of informative censoring due to the selective removal of a healthier group of transplant-eligible home HD patients. We performed a competing risk model analyses to take this into account. Our event of interest was all-cause mortality and the competing event was kidney transplantation in the home HD group and graft loss in the KTx group. In our study we used the Fine and Gray model,²⁰ which extends the Cox proportional hazards model to competing-risks data by considering the subdistribution hazard.

Effect modification by race was tested for the association of treatment modality with all-cause mortality. As the first-order interaction was statistically significant, we performed stratified analysis in African American and white patients. Statistical analyses were performed using Stata MP version 13 (Stata Corporation, College Station, TX, USA) and SAS, version 9.4 (SAS Institute, Cary, NC, USA).

Results

Baseline characteristics

Baseline characteristics of home HD and KTx patients before and after matching are shown in Table 1. Home HD patients were older, more likely to be male, diabetic, and white, had a higher prevalence of hypertension, atherosclerotic heart disease, congestive heart failure, and other cardiovascular disease and had higher serum albumin. After PS matching, all baseline variables were well balanced between home HD and KTx patients (Table 1).

Table 1.

Baseline characteristics of the unmatched and the 1:1 propensity score-matched cohort

	Unmatched			Matched
	Home HD	KTx	Std. Diff.	Home HD	KTx	Std. Diff.
	(n =2,830)	(n = 73,976)		(n=2,000)	(n=2,000)
Age (years)	53 ± 15	51 ± 13	0.129	54 ± 15	54 ± 14	−0.015
Female (%)	34	39	−0.091	35	35	0.016
Diabetes mellitus (%)	60	36	0.502	57	57	0
Race/Ethnicity (%)
Whites	70	51	0.125	70	72	−0.044
African-American	21	26	−0.127	19	19	−0.007
Asian	2	8	−0.263	3	2	0.039
Hispanic	5	13	−0.313	6	5	0.033
Other	2	2	−0.007	2	2	0.010
Primary insurance (%)
Medicare	40	19	0.614	38	40	−0.041
Medicaid	3	14	−0.282	4	4	0.008
Other	57	66	0.171	58	57	0.037
Comorbid States (%)
Alcohol abuse	0.2	0.7	−0.070	0.3	0.3	−0.008
History of cancer	4	2	0.126	4	4	0.010
Hypertension	72	61	0.238	74	74	0.008
Cerebrovascular disease	1	2	−0.066	2	2	−0.046
Artherosclerotic Heart Disease	26	6	0.591	21	21	−0.001
Congestive heart failure	49	6	1.089	36	40	−0.079
Other cardiovascular disease	22	5	0.524	18	20	−0.046
Chronic Obstructive Pulmonary Disease	6	1	0.253	4	4	0.005
Access Type at time of Home HD initiation/time of KTx (%)
AV Fistula	57	11	1.114	49	48	0.010
AV Graft	8	2	0.315	7	7	−0.006
CVC Catheter	19	31	−0.267	24	23	0.024
Other	0	0.6	−0.097	0.1	0	0.032
Unknown	15	56	−0.947	20	22	−0.034
Cause of ESRD (%)
Diabetes	35	31	0.101	37	37	0.002
Hypertension	22	20	0.042	21	21	0.002
Glomerulonephritis	18	19	−0.036	18	17	0.018
Cystic kidney disease	6	9	−0.107	7	7	−0.006
Other urologic reason	18	18	−0.006	18	18	−0.019
Unknown	0	2	−0.201	0	0	0
Laboratory Tests at time of Home HD initiation/time of KTx
Serum albumin (g/dL)	3.9 ± 0.5	3.5 ± 0.6	0.785	3.9 ± 0.5	3.9 ± 0.6	0.007
Blood hemoglobin (g/dL)	11.1 ± 1.3	10.3 ± 1.8	0.540	11.0 ±1.3	11.0 ± 1.6	0.034
Other
Total ESRD time before modality initiation (days)	387 ± 374	1012 ± 1065	−0.782	409 ± 383	396 ± 391	−0.035
Body Mass Index (kg/m2)	30 ± 7	28 ± 6	0.190	29 ± 7	29 ± 6	−0.001

Similar differences in baseline characteristics between home HD and KTx patients were seen in African American and white cohorts before matching, but became well balanced after PS matching (Table 2).

Table 2.

Baseline characteristics of the unmatched and the 1:1 propensity score-matched African Americans and Whites

	African Americans						Whites
	Unmatched			Matched			Unmatched			Matched
	Home HD	KTx	Std. Diff.	Home HD	KTx	Std. Diff.	Home HD	KTx	Std. Diff.	Home HD	KTx	Std. Diff.
	(n =585)	(n = 19,268)		(n=364)	(n=364)		(n =1,961)	(n = 46,884)		(n=1,447)	(n=1,447)
Age (years)	47 ± 13	50 ± 13	−0.241	49 ± 13	50 ± 13	−0.027	55 ± 14	52 ± 14	0.244	56 ± 14	56 ± 13	−0.036
Female (%)	33	41	−0.153	34	34	0.006	35	38	−0.057	35	34	0.032
Diabetes mellitus (%)	59	36	0.493	59	59	0	60	37	0.471	56	56	0
Primary insurance (%)
Medicare	34	17	0.614	31	31	−0.002	42	22	0.558	39	41	−0.041
Medicaid	4	19	−0.258	7	7	−0.022	3	12	−0.330	3	3	−0.011
Other	61	64	0.487	63	61	0.023	56	66	0.066	58	56	0.045
Comorbid States (%)
Alcohol abuse	0.2	0.5	−0.050	0.3	0	0.074	0.3	0.8	−0.077	0.3	0.3	0
History of cancer	4	1	0.186	4	2	0.082	4	2	0.116	5	5	−0.013
Hypertension	83	52	0.705	82	84	−0.058	68	66	0.070	72	72	−0.008
Cerebrovascular disease	0.2	2.1	−0.185	0.3	0	0.074	2	2	−0.039	2	3	−0.023
Artherosclerotic Heart Disease	24	4	0.623	16	19	−0.080	27	7	0.558	22	22	−0.010
Congestive heart failure	50	7	1.085	34	37	−0.063	49	7	1.084	37	40	−0.053
Other cardiovascular disease	20	3	0.532	13	24	−0.008	23	6	0.509	19	20	−0.017
Chronic Obstructive Pulmonary Disease	6	0.7	0.282	3	4	−0.015	6	1	0.252	5	5	−0.003
Access Type at time of Home HD initiation/time of KTx (%)
AV Fistula	57	10	1.168	46	46	0	56	12	1.042	48	48	0.007
AV Graft	13	2	0.411	13	13	−0.016	7	1	0.287	5	6	−0.018
CVC Catheter	16	32	−0.365	22	22	−0.013	20	32	−0.270	24	24	0.013
Other	0	0.6	−0.109	0	0	0	0.6	0.05	−0.093	0.07	0	0.037
Unknown	13	55	−1.001	19	18	0.028	16	54	−0.850	21	22	−0.013
Cause of ESRD (%)
Diabetes	31	30	0.024	37	36	0.023	36	32	0.094	37	35	0.032
Hypertension	37	35	0.039	32	36	−0.081	18	15	0.078	17	19	−0.038
Glomerulonephritis	20	17	0.084	20	19	0.021	17	20	−0.058	17	16	0.026
Cystic kidney disease	2	3	−0.060	3	2	0.054	8	12	−0.143	8	8	0.015
Other urologic reason	9	14	−0.133	8	7	0.041	21	21	−0.008	21	22	−0.035
Laboratory Tests at time of Home HD initiation/time of KTx
Serum albumin (g/dL)	4.0 ± 0.5	3.4 ± 0.6	1.082	3.9 ± 0.5	3.9 ± 0.5	0.016	3.9 ± 0.5	3.5 ± 0.6	0.659	3.8 ± 0.5	3.8± 0.6	−0.011
Blood hemoglobin (g/dL)	10.9 ± 1.3	9.9 ± 1.8	0.670	10.7 ±1.4	10.7 ± 1.6	−0.010	11.1 ± 1.3	10.4 ± 1.7	0.467	11.0 ±1.3	11.0 ± 1.7	0.039
Other
Total ESRD time before modality initiation (days)	485 ± 423	1489 ± 1171	−1.141	505 ± 449	509 ± 459	−0.010	350 ± 342	829 ± 951	−0.671	362 ± 357	379 ± 347	−0.048
Body Mass Index (kg/m2)	30 ± 7	29 ± 7	0.131	29 ± 7	29 ± 6	−0.034	30 ± 7	28 ± 6	0.205	29 ± 7	29 ± 6	0.009

Mortality in the entire PS matched cohort

Median follow-up time was 246 days (IQR: 99–365 days) for home HD patients and 845 days (IQR: 400–1,278 days) for KTx recipients. There were 261 deaths (13%, mortality rate 145; 95%CI: 128–164/1000 patient-years) in the home HD group, and 150 deaths (7.5%, 32; 95%CI: 27–38/1000 patient-years) in the KTx group. Figure 2 Panel A shows the probability of 5-year survival of home HD and KTx patients in the PS matched group. Home HD patients had a higher mortality risk compared to KTx patients (hazard ratio (HR): 4.06, 95% confidence interval (CI): 3.27–5.04) over the entire 5-year follow-up period (Table 3).

Figure 2.

Association between renal replacement type (home hemodialysis (Home HD) versus kidney transplantation (Kidney Tx)) and mortality using Kaplan-Meier curves in propensity score matched cohorts in all patients (Panel A), African Americans (Panel B) and Whites (Panel C)

Table 3.

Mortality risk of home hemodialysis patients compared to kidney transplant recipients in group of patients with different donor characteristics using propensity score matched cohorts in the first year and thereafter

		Entire follow-up period			0–365 days			>365 days
		Number of Patients/Events/% of Events	HR	95% CI of HR	Number of Events/% of Events	HR	95% CI of HR	Number of Events/% of Events	HR	95% CI of HR
All donors	All patients	4,000/411/10.3%	4.06	3.27–5.04	264/6.6%	3.77	2.86–4.97	147/3.7%	4.53	3.23–6.34
	African American	728/55/7.6%	2.06	1.16–3.67	29/4.0%	1.62	0.77–3.39	26/3.6%	2.84	1.23–6.55
	White	2,894/332/11.5%	4.38	3.43–5.59	224/7.7%	4.21	3.10–5.73	108/3.8%	4.66	3.14–6.92
Living donors - All patients		3,262/308/9.4%	6.45	4.93–8.43	202/6.2%	7.56	5.15–11.09	106/3.2%	5.42	3.66–8.01
Deceased donors - All patients		3,354/375/11.2%	3.45	2.75–4.31	261/7.8%	2.85	2.20–3.71	114/3.4%	5.16	3.50–7.60
ECD donors - All patients		2,212/287/13.0%	3.90	3.00–5.07	202/9.1%	3.38	2.48–4.60	85/3.9%	5.25	3.36–8.22
SCD donors - All patients		3,810/378/9.9%	4.64	3.69–5.84	249/6.5%	4.25	3.17–5.69	129/3.4%	5.30	3.69–7.60
DCD donors - All patients		1,702/198/11.6%	3.70	2.75–5.21	152/8.9%	3.05	2.15–4.32	46/2.7%	7.67	4.06–14.51
Non DCD donors - All patients		3,240/357/11.0%	3.40	2.70–4.27	248/7.7%	3.00	2.28–3.93	109/3.3%	4.42	2.98–6.56
KDPI>20 - All patients		3,176/344/10.8%	4.30	3.38–5.47	241/7.6%	3.80	2.85–5.07	103/3.2%	5.46	3.64–8.19
KDPI≤20 - All patients		1,940/205/10.6%	5.40	3.89–7.49	144/7.4%	5.04	3.38–7.51	61/3.2%	6.14	3.55–10.61

Associations of home HD vs. KTx with higher risk of mortality were found in both African Americans (Figure 2 Panel B) and whites (Figure 2 Panel C). There was a significant interaction with race for the association of treatment modality with all-cause mortality (p=0.018): mortality risk was over 4-fold higher (HR: 4.38, 95%CI: 3.43–5.59) in whites but 2-fold higher (HR: 2.06, 95%CI: 1.16–3.67) in African Americans (Figure 3). In African Americans, mortality risk increased after the first year as the survival lines were separated only after this timepoint (p<0.001 for interaction with time), while in whites the survival lines were separated from the beginning of the follow-up. In addition, there was no difference between home HD and KTx for first-year mortality risk in African American patients (HR: 1.62, 95%CI: 0.77–3.39), while white home HD patients had higher mortality risk (HR: 4.21, 95%CI: 3.10–5.73) even during the first year (Table 3 and Table S19).

Figure 3.

Mortality risk of home hemodialysis patients compared to kidney transplant recipients in group of patients with different donor characteristics using propensity score matched cohorts

Similar results were found when we used alternative censoring in our sensitivity analyses (Figures S2–S3) or competing risk regression analyses (Table S18).

Mortality in the PS matched subcohorts matching by different type of KTx donors

Baseline characteristics of subcohorts comparing home HD patients to KTx patients by different type of KTx donors, before and after matching are shown in Tables S1–S16. Significant differences in baseline characteristics of home HD and KTx patients were detected before matching, however, after PS matching all baseline variables became well balanced. Home HD patients had higher mortality risk compared to KTx patients in all PS matched subcohorts Figures S4–S11. Similar associations were found among African Americans and whites in all subcohorts, and there was effect modification by race, with a higher mortality risk observed in whites. Additionally, in most of the subcohorts (deceased-, SCD-, non-DCD-, KDPI>20 or KDPI≤20 donors) there was no difference between African American home HD and KTx patients in first-year mortality risk, while white home HD patients had higher mortality risk than their KTx counterparts even during the first year.

Discussion

In this contemporary cohort of incident home HD patients and KTx recipients in the United States, we examined the association between type of renal replacement modality with all-cause mortality. Patients who received KTx had significantly better survival regardless of the type of kidney donor type. In addition, we also detected an interaction between race and renal replacement type for mortality outcomes, in that KTx had a stronger protective effect on survival in white recipients. In contrast, home HD was associated with similar first year survival as KTx in African Americans.

Our results confirm that KTx is the treatment of choice for end stage renal disease, even when compared to home HD. Home HD patients had 4-times higher risk of mortality compared to their KTx counterparts. These findings are similar to those observed in studies examining nocturnal home HD from Ontario, Canada.¹⁴ Conversely, results from previous studies have shown benefits of home HD and may have led to the hypothesis that home HD patients had similar survival to KTx patients. Home HD may provide certain advantages in ESRD patients, including improved blood pressure control^21–24 despite increased extracellular fluid volume.^21,25 Nocturnal home HD patients have shown reductions in total peripheral resistance and plasma norepinephrine levels.²⁶ In addition, randomized trials and meta-analyses suggest that home HD is associated with reduction in left ventricular mass compared to conventioal in-center HD.^21,24,27,28 Home HD treatment improves anemia without altering erythropoietin requirements or iron status.²⁹ Additonally, home HD provides better phosphorus control than in-center HD.^24,30–32 Sleep disorders such as sleep apnea and restless legs syndrome, which are predictors of mortality in ESRD patients,^33,34 improve in home HD patients,³⁵ but are still very prevalent in KTx population.^36,37 Conversion from in-center to home HD reduces the number of apnea or hypopnea events in patients with obstructive sleep apnea.³⁵ However, despite these previous findings, our data confirm that regardless of the type of kidney donor, KTx provides better survival than home HD for patients with ESRD.

Interestingly, clinically and statistically significant interactions were found between race (African American and white patients) and renal replacement therapy type for mortality outcomes. Among whites, home HD patients had significantly higher mortality risk in the first year and thereafter. However, the first-year survival was similar with home HD and KTx in African Americans except when the kidney was donated from a living donor, and the mortality risk associated with home HD increased only after the first year. This finding should be interpreted with caution. The subgroup of African Americans in our study was relatively small (n=728), consequently the observed event number was also small (n=55), so this subgroup analysis may be underpowered. Further studies are needed to confirm or reject our results. These results may suggest that while waiting for a transplant all patients requiring renal replacement therapy could benefit from home HD. Furthermore, it is possible that home HD is the best dialysis option for African American patients, although our study didn't compare it directly with conventional in-centre hemodialysis or with peritoneal dialysis, and hence this may need further examination in future studies. There are several potential explanations for the lack of first-year survival benefit in KTx living donor versus home HD in African American patients. First, African Americans compared to whites have a higher probability of developing ESRD, secondary to genetic predisposition even in strata of patients with a history of diabetes and hypertension.³⁸ In addition, the patient and renal allograft survival is significantly lower in African Americans than in whites,^15,39,40 which might be due to higher immunologic risk, lower medication adherence, or decreased access to pre- and posttransplantation care.^41–43 African American patients on maintenance dialysis also have better survival rates than whites⁴⁴ secondary to several factors such as higher muscle mass,⁴⁵ and better nutritional and inflammatory status.⁴⁶ A similar survival advantage can be detected among home HD patients. In a recent study, we examined patients who initiated maintenance dialysis between 2007 and 2011 in any of 2217 dialysis facilities operated by a single large dialysis organization, with follow-up through Dec. 31, 2011. Compared to whites, African Americans undergoing home HD had a lower risk for death even after adjustment for important confounders.⁴⁷ Such discrepant associations of African American race with survival in in home HD vs. in KTx could explain the effect modificaiton by race described in our study. Higher survival chance in home HD and lower survival in KTx and the economic and social inequality between whites and African Americans can also explain the result of this study.

Our study has number of strengths. It is the first comparison of mortality for home HD patients and KTx recipients from the United States. This is the first study using a PS matched approach to balance measured confounders. Moreover, we were able to compare the mortality risk of home HD with KTx from different types of kidney donors. In addition, we performed sensitivity analyses with an alternate censoring method by continuing to follow patients after home HD therapy ended, which confirmed our results. Furthermore, we also performed sensitivity analyses using competing risk regression analyses to take into account informative censoring due to the selective removal of a healthier group of transplant eligible home HD patients. The results of these analyses were qualitatively similar to our main results. Finally, we assessed the effect modification of race in the association of modality type with the mortality outcome.

The results of our study should be interpreted in light of some potential limitations. First, home HD data were derived from facilities operated by a single dialysis provider. However, this constitutes almost one-third of all patients undergoing maintenance dialysis in the US. Second, we acknowledge that our sample size and event numbers in home HD patients are small. The relatively small sample size may have a bearing on the proportion of African American patients (n=728) given small number of events (n=55), where the effect size exhibited large variability, i.e., as low as 16% but as high as 367% for higher mortality risk. Hence, our result should be qualified in this context and need to be confirmed in larger studies involving more African American patients and more events. However, to the best of our knowledge ours is still the largest cohort of home HD patients assembled to date and compares outcomes to a large cohort of KTx recipients. In addition, our result may not be applicable to populations outside the US, as the nocturnal home hemodialysis practice is significantly different in Canada or Europe. Furthermore, median follow-up time in home HD patients was relatively short. The main reason for this was that our home HD patients were transferred to other dialysis modalities after a relatively short period of time. Further studies are needed to identify the cause of this phenomenon. Third, we did not have data regarding the home HD patients’ waitlist status, consequently we were not able to perform subgroup analysis in this subcohort. Lastly, despite the fact that we were able to PS-match our cohorts for many confounding factors, there are likely remaining unmeasured or unknown confounders that could have affected the results of this study.

Conclusion

In conclusion, patients who received KTx had significantly better survival compared to home HD patients, regardless of kidney donor type. African American home HD patients had similar first year survival to African American KTx patients without a living donor. Further studies are needed to confirm and understand the reasons underlying racial differenes in mortality risk in home HD versus KTx patients.

List of abbreviations

CI

confidence interval

DCD

donor after cardiac death

ECD

extended criterion donor

ESRD

end stage renal disease

HHD

home hemodialysis

HD

hemodialysis

HR

hazard ratios

IQR

interquartile range

KDPI

Kidney Donor Profile Index

KTx

kidney transplant

PS

propensity score

SD

standard deviation

SCD

standard criterion donor

US

United States