Abstract
Rationale & Objective
Transportation insecurity is a social risk factor of particular importance to individuals with end-stage kidney disease (ESKD), as most individuals need to travel multiple times a week to dialysis treatment. Advancing home modalities for individuals with ESKD experiencing transportation insecurity may be beneficial by reducing travel burden and improving access.
Study Design
Retrospective cohort study.
Setting & Participants
Individuals with ESKD treated with in-center hemodialysis (HD) at a large, national dialysis organization.
Exposures
The main transportation mode to HD is categorized into private transportation (individuals who drive themselves or have a family member/friend drive) or those who lack private transportation (Medicaid non-emergency medical transportation, paratransit, public transportation, private pay non-emergency medical transportation, and other).
Outcomes
Transition to home dialysis is defined as an individual who has completed at least 1 training treatment for home therapies or at least 1 dialysis treatment at home.
Analytic Approach
Log-binomial multivariate regression models to estimate adjusted incidence rate ratios of home dialysis transition by transportation mode.
Results
Individuals who lacked private transportation were significantly less likely to transition to home dialysis compared with those who drove themselves or had a family member/friend drive them to HD. Adjusted incidence rate ratios for home dialysis transition were 47%-58% lower in nonprivate transportation groups compared with those with private transportation, ranging from 0.42 in individuals relying on Medicaid transportation benefits (95% confidence interval, 0.35-0.50; P < 0.001) to 0.53 (95% confidence interval, 0.41-0.67; P < 0.001) among paratransit users.
Limitations
Single transportation assessment, exclusion of individuals already on home dialysis, and absence of caregiver data.
Conclusions
Individuals with ESKD receiving in-center HD who lack private transportation may have reduced access to home dialysis, even though this group may benefit from home modalities. Better identifying transportation barriers and targeting home modalities for those with transportation insecurity may reduce the adverse consequences of missed dialysis related to transportation barriers and be an additional opportunity to increase home dialysis uptake.
Index Words: End-stage kidney disease, dialysis, home modalities, transportation, social risk
Plain-Language Summary
Transportation is a key barrier for many individuals receiving in-center dialysis care. Nonetheless, the majority of individuals in the United States receive their dialysis treatment at an in-center facility. In a study of patients with end-stage kidney disease treated at in-center dialysis facilities, we examined the association between mode of transportation to dialysis and transition to home dialysis. We found that individuals who do not drive themselves or have a family member or friend drive them to dialysis were less likely to transition to home dialysis in the follow-up period. Our findings raise policy opportunities to support individuals who may face transportation challenges with ways to receive dialysis at home and reduce their transportation needs.
Home dialysis provides individuals with end-stage kidney disease (ESKD) treated with dialysis with increased flexibility and independence.1 Yet of the approximately 550,000 individuals in the United States with ESKD treated with dialysis, only 14.5% of prevalent dialysis patients performed home dialysis in 2022.2 This is in contrast to other countries, such as Australia and New Zealand, where a quarter or more of prevalent dialysis patients use home modalities.3,4
The 2019 Executive Order on Advancing American Kidney Health emphasized increased home dialysis as a key goal in national efforts to improve kidney care.5 Responding to this order, the Centers for Medicare and Medicaid Services launched the mandatory ESRD Treatment Choices (ETC) model that integrated home dialysis utilization, as well as kidney transplantation waitlisting, as a key target for dialysis facilities and managing clinicians through financial incentives.6 An initial report demonstrated small but significant increases in home dialysis utilization among incident patients in ETC markets, but there was no significant change among Medicare fee-for-service beneficiaries.7 Analyses contracted by the Centers for Medicare and Medicaid Services have not demonstrated differences in the rate of change in the use of home dialysis among Medicare beneficiaries between ETC and non-ETC markets.8
Financial incentives for home dialysis remain one piece of the complex system of providing dialysis care in the United States. Increasing attention has turned to the role of social risks in health and health outcomes.9,10 Although prior policies have implemented screening for social drivers of health as quality measures, less attention has focused on the link between how social risks may influence modality decisions.11,12
Transportation insecurity is a social risk factor of particular importance to people on dialysis. Individuals with ESKD treated with in-center dialysis need to travel to dialysis facilities multiple times a week. As such, advancing home modalities for individuals with transportation insecurity may be particularly beneficial by reducing travel burden. However, current understanding of the association of transportation access and home dialysis uptake remains limited. Among a cohort of patients receiving in-center dialysis, we examined the association between transportation mode and transition from in-center hemodialysis (HD) to home dialysis (peritoneal dialysis and HD). We defined individuals as transitioning to home dialysis as those who completed at least 1 training treatment for home therapies or at least 1 dialysis treatment at home. Studying the relationship between transportation mode and home dialysis transition is an initial step in understanding how transportation is associated with important clinical outcomes.
Methods
Study Design, Setting, and Participants
We conducted a retrospective cohort study using data obtained from a large dialysis organization. The study included all adult patients (age ≥18 years) with ESKD on April 1, 2022, who had already been established for at least 90 days as receiving in-center HD and who had already completed at least 1 transportation assessment within the year before that date. For adults meeting these eligibility criteria, we ascertained the study outcome (transition to home dialysis) and time-varying dialysis utilization covariates during discrete 30-day periods for up to 1 year (360 days) of follow-up from April 1, 2022, to March 31, 2023. Patients were censored because of death, kidney transplantation, or discharge from the dialysis provider (eg, recovery of kidney function, transfer to another provider, or loss to follow-up). We followed STROBE reporting guidelines (see Table S1). The UC Davis Institutional Review Board approved the study.
A Community Advisory Board (CAB) was established in the fall of 2022 at the launch of the study. The Community Advisory Board has met nearly every other month since 2022 with members of the study team (NR and BBH) to provide feedback on all steps of the research study, including design, analysis, interpretation of findings, and goals for dissemination. During the key analysis of the presented data, 5 community advisory board members were actively involved: 2 of these individuals were currently on dialysis, and 3 had been on dialysis and had received a kidney transplant. Collectively, community advisory board members had experience with home modalities, including peritoneal dialysis. They represent diversity in terms of age, geography, racial and ethnic identity, and transportation mode utilization.
Exposure
The dialysis organization implemented the transportation assessment across the organization, and the assessment was completed by staff. The assessment documented whether a patient had private transportation (yes or no). If a patient did not have private transportation, the assessment then included information on the main transportation source to dialysis, with the following response options: public transit, paratransit, Medicaid, or private pay non-emergency medical transportation (NEMT). We used the transportation assessment result closest to but not after the start of follow-up, April 1, 2022.
Outcome of Transition to Home Dialysis
During follow-up, we ascertained whether patients transitioned to home dialysis (either home HD or peritoneal dialysis). The transition to home dialysis was identified as completing at least 1 home therapy training treatment or dialysis treatment at home. We operationalized outcomes based on standard definitions and those used by the large dialysis organization.
Covariates
The following patient covariates were determined from the dialysis organization’s electronic health record and based on information as of the date of cohort entry: age, biological sex, race and ethnicity (understood as social constructs and included since transportation access has been shaped by racism), median household income and education (based on residential ZIP code), patient-level geographic categorization using RUCA (rural-urban commuting area) codes, dialysis facility-level measure of area deprivation (Social Deprivation Index), comorbidities (diabetes mellitus, hypertension, heart failure, dementia), cause of ESKD, and time since the onset of ESKD requiring dialysis. Median household education was calculated using 5-digit ZCTA linked to the 2016-2020 American Community Survey and was modeled as a continuous variable of the percentage of the population 25 years and over with a high school degree (or equivalent).13 Median household income was calculated from the 2021 American Community Survey 1-Year Estimates.14 The research team provided the dialysis organization with a crosswalk to link patient zip codes to ZCTA education and income variables.
For every 30 days, we calculated the total expected HD sessions and the total missed HD sessions, excluding those related to hospitalization or emergency department visits. For every 30 days, we obtained data on total expected HD sessions and total missed dialysis treatments from the dialysis organization’s electronic health record. All-cause missed dialysis treatment was operationalized to include all missed HD treatments that were not rescheduled or missed because of hospitalization, including those caused by transportation or any other cause. Time-varying covariates included (1) the duration of time that elapsed from the patient’s first date of dialysis for ESKD to the start of each person-period and (2) the log-transformed missed dialysis session rate for the individual in the previous person-period. We included these 2 covariates because prior research suggests that modality decision occurs early on with the initiation of dialysis, and missed dialysis appointments may be associated with the likelihood of referral to home dialysis modalities.2 We log-transformed the missed dialysis session rate to be on the same logarithmic scale as the expected value of the response variable in our generalized estimating equation models, which used a log link function.
Analytical Approach
Analyses were conducted using Stata/SE Version 15. We performed descriptive analysis to characterize the study population by transportation mode. We used log-binomial multivariate generalized estimating equations regression models to estimate adjusted incidence rate/risk ratios (aIRRs) to study home dialysis transition by transportation mode and covariates. The units of analysis for all regression analyses were discrete patient-months (30-day periods). These models use the natural logarithm function as the link function, so the inverse natural logarithm function was applied to estimate regression coefficients to yield aIRRs and 95% confidence intervals (CIs). To account for nesting of follow-up months within patients and for slight model misspecification regarding the outcome distribution assumptions, we used cluster-robust standard errors. Hypothesis tests were 2-sided with α = 0.05.
Results
The overall cohort included 115,982 individuals with ESKD receiving in-center HD, with an average age of 63.4 (±standard deviation, 14.0), 42.6% prevalence of females, 36.6% prevalence of Black individuals, and 25.8% residing outside of the urban core (10.6% large rural, 7.9% suburban, and 6.8% small town/rural). See Table 1 for additional descriptive data of the cohort. Nearly a third of patients (28.7%) did not have private transportation. Among the entire cohort, 14.6% relied on Medicaid transportation benefits, 7.6% used paratransit, 4.5% traveled by public transportation, 1.9% used private pay non-emergency medical transportation, and 39 individuals had an unspecified mode of non-private transportation. The median time since dialysis initiation in the cohort was 3.6 years (25% quartile 1.7 years; 75% quartile 6.6 years).
Table 1.
Baseline Cohort Characteristics
| Charcateristic | Total Cohort N=115,982 (100%) |
Private Transportation N=82,718 (71.3%) |
Nonprivate Transportationa N=33,264 (28.7%) |
||||
|---|---|---|---|---|---|---|---|
| Medicaid N=16,981 (14.6%) |
Paratransit N=8,795 (7.6%) |
Public Transit N=5,244 (4.5%) |
Private Pay NEMT N=2,205 (1.9%) |
Other/unknown N=39 (0.03%) |
|||
| Age, y | 63.4 (14.0) | 63.2 (14.2) | 61.6 (13.6) | 66.8 (12.5) | 64.7 (13.05) | 69.3 (11.9) | 63.6 (14.5) |
| Median household income ($)b | 58,579 (23,885) | 59,713 (23,771) | 54,303 (23,443) | 58,233 (25,226) | 56,019 (21,579) | 56,840 (26,788) | 49,274 (17,047) |
| Adult households with high school education or greaterc | 33,954 (29.4) | 24,216 (29.5) | 4 ,987 (29.4) | 2,524 (28.9) | 1,588 (30.5) | 627 (28.6) | 12 (30.3) |
| SDId | 61.2 (26.5) | 60.6 (26.5) | 63.6 (26.2) | 61.7 (26.9) | 62.6 (25.7) | 59.6 (26.4) | 73.7 (18.8) |
| Time since the dialysis initiation, ye | 3.6 (1.7, 6.6) | 3.4 (1.6, 6.2) | 4.2 (2.1, 7.6) | 4.1 (2.1, 7.4) | 4.4 (2.2, 7.6) | 3.5 (1.7, 6.6) | 4.9 (3.1, 8.5) |
| Sex | |||||||
| Female | 49,366 (42.6) | 32,908 (39.8) | 8,554 (50.4) | 4,336 (49.6) | 2,447 (46.7) | 1,074 (48.7) | 17 (43.6) |
| Race | |||||||
| American Indian/Alaska Native | 861 (0.7) | 581 (0.7) | 131 (0.8) | 74 (0.8) | 65 (1.2) | <30 | - |
| Asian | 4,299 (3.7) | 3,191 (3.9) | 625 (3.7) | 314 (3.6) | 128 (2.4) | ≥30 | - |
| Black | 42,476 (36.6) | 28,230 (34.1) | 7,482 (44.1) | 3,719 (42.3) | 2,252 (42.9) | 775 (35.2) | 18 (46.2) |
| Native Hawaiian/Pacific Islander | 1,922 (1.7) | 1,397 (1.7) | 252 (1.5) | 156 (1.8) | 98 (1.9) | <30 | - |
| White | 65,171 (56.2) | 48,308 (58.4) | 8,372 (49.3) | 4,469 (50.8) | 2,665 (50.8) | 1,339 (60.7) | 18 (46.2) |
| Missing race | 1,253 (1.1) | 1,011 (1.2) | 119 (0.7) | 63 (0.7) | 36 (0.7) | <30 | - |
| Ethnicity | |||||||
| Hispanic | 21,158 (18.2) | 14,576 (17.6) | 3,770 (22.2) | 1,628 (18.5) | 690 (13.2) | ≥30 | - |
| Non-Hispanic | 93,423 (80.5) | 67,014 (81.0) | 13,078 (77.0) | 7,095 (80.7) | 4,512 (86.0) | 1,693 (76.8) | 31 (79.5) |
| Missing | 1,401 (1.2) | 1,128 (1.4) | 133 (0.8) | 72 (0.8) | 42 (0.8) | <30 | - |
| RUCAf | |||||||
| Large rural | 12,259 (10.6) | 9,104 (11.0) | 1,547 (9.1) | 570 (6.5) | 817 (15.6) | ≥30 | - |
| Small town/rural | 7,937 (6.8) | 5,842 (7.1) | 1,211 (7.1) | 341 (3.9) | 453 (8.6) | ≥30 | - |
| Suburban | 9,188 (7.9) | 7,145 (8.6) | 1,007 (5.9) | 439 (5.0) | 430 (8.2) | ≥30 | - |
| Urban core | 86,003 (74.2) | 60,130 (72.7) | 13,213 (77.8) | 7,397 (84.1) | 3,506 (66.9) | 1,724 (78.2) | 33 (84.6) |
| Comorbidities | |||||||
| Heart failure | 22,939 (19.8) | 15,684 (19.0) | 3,605 (21.2) | 1,925 (21.9) | 1,178 (22.5) | ≥30 | - |
| Diabetesg | 71,490 (61.6) | 49,038 (59.3) | 11,314 (66.6) | 6,092 (69.3) | 3,413 (65.1) | 1,608 (72.9) | 25 (64.1) |
| Hypertensionh | 92,984 (80.2) | 66,489 (80.4) | 13,504 (79.5) | 6,977 (79.3) | 4,258 (81.2) | 1,722 (78.1) | 34 (87.2) |
| Dementia | 1,528 (1.3) | 864 (1.0) | 300 (1.8) | 181 (2.1) | 70 (1.3) | ≥30 | - |
| Cause of ESKD | |||||||
| Diabetes | 56,154 (48.8) | 38,181 (46.5) | 9,077 (53.8) | 4,865 (55.7) | 2,709 (52.1) | 1,303 (59.6) | 19 (48.7) |
| HTN/large vessel disease | 37,564 (32.6) | 27,336 (33.3) | 5,144 (30.5) | 2,723 (31.2) | 1,713 (32.9) | 633 (29.0) | 15 (38.5) |
| Otheri | 21,429 (18.6) | 16,599 (20.2) | 2,649 (15.7) | 1,147 (13.1) | 779 (15.0) | ≥30 | - |
Data are presented as mean (±standard deviation) or n (%), unless otherwise listed below.
Table adapted from Razon et al.16
Abbreviations: ESKD, end-stage Kidney Disease; GED, general educational development; HTN, hypertension; NEMT, non-emergency medical transportation; RUCA, rural-urban commuting area; SDI, Social Deprivation Index.
Exact counts for small cell sizes for categorical variables are not reported to preserve confidentiality. A cell with <11 individuals is reported as “-.” To prevent the determination of those cell counts by referring to other cell counts in the row, we selected certain cell counts to report as either <30 or as 30+.
Median household income. Total = 115,061, missing 0.8%. Total: private pay, N= 81,981; Medicaid, N=16,933; paratransit, N=8,728; public transit, N=5,196; private pay NEMT, N=2,184; other/missing. N=39.
Adult household with a GED or greater. Total = 115,352, missing 0.5%. Total: private Pay, N = 82,198; Medicaid, N = 16,973; paratransit, N = 8,745; public transit, N = 5,204; private pay NEMT, N = 2,193; other/missing, N = 39.
Dialysis facility SDI based on zip code. Total = 111,784, missing 3.6%. Total: private pay, N = 80,538; Medicaid, N = 15,857; paratransit, N = 8,284; public transit, N = 5,162; private pay, NEMT N = 1,904; other/missing, N = 39.
Total time on dialysis prior to start of study. Total = 115,982. None missing. Listed as median (25% quartile, 75% quartile) Total: Private Pay N = 82,718; Medicaid N = 16,981; Paratransit N = 8,795; Public transit N = 5,244; Private pay NEMT N = 2205; Other/Missing N = 39.
Rural-urban commuting area codes (RUCA): Urban core (RUCA, 1); suburban (RUCA 2-3); large rural (RUCA, 4-6); small town/rural (RUCA, 7-10). Washington State Department of Health. (2016). Guidelines for using Rural-Urban Classification Systems for Community Health Assessment. Available at: https://doh.wa.gov/data-and-statistical-reports/data-guidelines. Total = 115,387, missing 0.5%.
Defined as the sum of diabetes uncomplicated, diabetes complicated, and diabetes as a cause of ESKD.
Defined as the sum of hypertension uncomplicated, hypertension complicated, and hypertension and large vessel disease as a cause of ESKD.
Total individuals with cause of ESKD listed =115,147, missing 0.72%. Other causes of ESKD include: cystic/hereditary/congenital, glomerulonephritis, interstitial nephritis/pyelonephritis, neoplasm/tumor, secondary glomerulonephritis/vasculitis, and transplant complications.
Association of Transportation With Transition to Home Dialysis
During the 12-month follow-up, the unadjusted rate of home dialysis transition was 2.1 per 100 person-years. Compared with private transport, all other modes of transportation were associated with lower adjusted rates of home dialysis transition (Tables 2 and 3). Adjusted IRRs for home dialysis transition were an estimated 47%-58% lower in nonprivate transportation groups compared with those with private transportation, ranging from 0.42 in individuals relying on Medicaid transportation benefits (95% CI, 0.35-0.50; P < 0.001) to 0.53 (95% CI, 0.41-0.67; P < 0.001) among paratransit users. American Indian and Alaskan Natives had the highest rate of home dialysis transition, although this was not statistically significant (aIRR, 1.23; 95% CI, 0.82-1.83; P = 0.32). Hispanic, Asian, and Black patients had statistically lower rates of home dialysis transition compared with White patients ([aIRR, 0.59; 95% CI 0.51-0.68; P < 0.001], [aIRR, 0.67; 95% CI, 0.50-0.89; P = 0.006], and [aIRR, 0.86; 95% CI, 0.77-0.95; P = 0.003], respectively).
Table 2.
Transition to Home Dialysis
| Variable | No. of Events/Person-year (Unadjusted Rate of Home Dialysis Transition Per 100 Person-years) | aIRR (95% CI) |
|---|---|---|
| Transportation mode | ||
| Private transport | 1,924/75,979 (2.53) | 1.0 (reference) |
| Medicaid | 138/14,901 (0.93) | 0.42b(0.35-0.50) |
| Paratransit | 73/7,725 (0.94) | 0.53b (0.41-0.67) |
| Private pay NEMT | 15/1,661 (0.90) | 0.47a (0.27-0.81) |
| Public transit | 48/4,890 (0.98) | 0.47b(0.35-0.63) |
| Race/ethnicity | ||
| White | 1,007/40,814 (2.47) | 1.0 (reference) |
| AIAN | 27/780 (3.46) | 1.23 (0.82-1.83) |
| Asian | 53/3,852 (1.38) | 0.67a (0.50-0.89) |
| Black | 782/40,065 (1.95) | 0.86a (0.77-0.95) |
| Hispanic | 277/17,316 (1.60) | 0.59b (0.51-0.68) |
| NH/PI | 20/1,230 (1.63) | 0.74 (0.47-1.16) |
| Rurality | ||
| Urban | 1,455/77,750 (1.87) | 1.0 (reference) |
| Large rural | 261/11,484 (2.27) | 1.15a (1.00-1.33) |
| Small town/rural | 220/7,403 (2.97) | 1.41b (1.20-1.65) |
| Suburban | 262/8,533 (3.07) | 1.44b(1.24-1.66) |
| Time since the dialysis initiation | ||
| 3-<9 mo | 313/2,220 (14.10) | 7.66b (6.55-8.96) |
| 9-<14 mo | 278/4,994 (5.57) | 3.39b (2.93-3.93) |
| 14 mo to <2 y | 447/15,237 (2.93) | 1.89b(1.67-2.13) |
| 2-<7 y | 856/55,852 (1.53) | 1.0 (reference) |
| 7-<12 y | 188/17,207 (1.09) | 0.72b (0.61-0.85) |
| 12-<17 y | 61/5,951 (1.03) | 0.66a (0.51-0.87) |
| ≥17 y | 55/3,733 (1.47) | 0.83 (0.62-1.11) |
Abbreviations: AIAN, American Indian/Alaska Native; aIRR, adjusted incident rate ratio; NEMT, non-emergency medical transportation; NH/PI, Native Hawaiian/Pacific Islander; RUCA, rural urban commuting area; SDI, social deprivation index.
Model using the generalized estimating equation log binomial function on a person-period data set with up to 13 person-periods per patient. Model also adjusted for: age, sex, comorbidities (congestive heart failure, diabetes, hypertension, and dementia), SDI, and missed appointment rate. The time since dialysis initiation covariate was time-varying and based on the patient’s time since dialysis initiation as of the start of each person’s period. See Table 3 for the full model and complete P values.
P < 0.005,
P < 0.001.
Table 3.
Full Log Binomial Regression Results of Association Between Transportation Mode and Home Dialysis Transition (1,148,616 30-day Observation Periods, 109,272 Persons)
| Variable | aIRR | 95% Confidence Interval | P Value |
|---|---|---|---|
| Age, y | 0.97 | 0.96-0.97 | < 0.001 |
| Female sex | 1.09 | 0.99-1.19 | 0.07 |
| Race (reference: White) | |||
| American Indian/Alaska Native | 1.23 | 0.82-1.83 | 0.32 |
| Asian | 0.67 | 0.50-0.89 | 0.006 |
| Black | 0.86 | 0.77-0.95 | 0.003 |
| Hispanic | 0.59 | 0.51-0.68 | < 0.001 |
| Native Hawaiian/other Pacific Islander | 0.74 | 0.47-1.16 | 0.19 |
| Missing/unknown | 0.40 | 0.27-0.61 | < 0.001 |
| RUCA (reference: urban) | |||
| Large rural | 1.15 | 1.00-1.33 | 0.049 |
| Small town/rural | 1.41 | 1.20-1.65 | < 0.001 |
| Suburban | 1.44 | 1.24-1.66 | < 0.001 |
| Binary logarithm of missed dialysis treatment ratea | 1.03 | 1.01-1.05 | 0.015 |
| Social deprivation index (in 20-point units)b | 0.41 | 0.20-0.84 | 0.014 |
| Comorbidities | |||
| Heart failure | 1.01 | 0.90-1.14 | 0.81 |
| Dementia | 0.56 | 0.28-1.13 | 0.10 |
| Diabetes | 0.94 | 0.85-1.03 | 0.19 |
| Hypertension | 1.11 | 0.99-1.25 | 0.07 |
| Dialysis duration (reference: 2-7 y) | |||
| 3-<9 mo | 7.66 | 6.55-8.96 | < 0.001 |
| 9-<14 mo | 3.39 | 2.93-3.93 | < 0.001 |
| 14 mo to <2 y | 1.89 | 1.67-2.13 | < 0.001 |
| 7-<12 y | 0.72 | 0.61-0.85 | < 0.001 |
| 12-<17 y | 0.66 | 0.51-0.87 | 0.003 |
| ≥17 y | 0.83 | 0.62-1.11 | 0.21 |
| Transportation mode (reference: private transportation) | |||
| Medicaid | 0.42 | 0.35-0.50 | < 0.001 |
| Paratransit | 0.53 | 0.41-0.67 | < 0.001 |
| Private pay NEMT | 0.47 | 0.27-0.81 | 0.006 |
| Public transit | 0.47 | 0.35-0.63 | < 0.001 |
Abbreviations: IRR, incident rate ratio; NEMT, non-emergency medical transportation; RUCA, rural urban commuting area; SDI, social deprivation index.
The missed dialysis treatment rate is primarily based on the number of missed dialysis treatments divided by the number of expected dialysis treatments during the study period. In the regression model, the binary logarithm of the missed dialysis treatment rate is entered. As a result, the IRR can be interpreted as describing the adjusted relative rate ratio when comparing two patients where the first patient has twice the missed treatments as the second. Said differently, for example, a home dialysis transition IRR of 1.2 means that each doubling in the missed dialysis percentage corresponds to an approximately 20 percent relative increase in home dialysis transition.
The coefficient describes the IRR associated when holding all things constant when switching between the neighborhoods containing one facility to another, where the difference between one facility and another is 20-point difference.
Association of Other Factors With Home Dialysis Uptake
Other factors associated with increased transition to home dialysis included residing in a nonurban setting and shorter duration of ESKD (Tables 2 and 3). In regard to geography, the relative rate of home dialysis transition was higher among patients living outside the urban core, ranging from an aIRR of 1.15 (95% CI, 1.00-1.33; P = 0.049) among those living in large rural communities to 1.45 (95% CI, 1.24-1.66; P < 0.001) for patients living in suburban areas.
In the adjusted model, those with the least time since starting dialysis had the highest home dialysis transition with a rapid decline to a more steady state of transition after 2 years (3-<9 months: aIRR, 7.66; 95% CI, 6.55-8.96; P < 0.001; 9-<14 months: aIRR, 3.39; 95% CI, 2.92-3.92; P < 0.001; 14 months to 2 years: aIRR, 1.89; 95% CI, 1.67-2.13; P < 0.001; 7-12 years: aIRR, 0.72; 95% CI, 0.61-0.85; P < 0.001), compared with the reference category of 2-7 years duration on dialysis.
Discussion
Facilitating transition to home dialysis modalities, including both peritoneal and home HD, has been a policy priority in the United States. In this large retrospective cohort study, we identified non-private transportation modes as important factors associated with low home dialysis uptake. This may be a potentially modifiable risk factor to improve home modality utilization. Individuals without private transportation were 47%-58% less likely to transition to home dialysis, defined as starting home dialysis training or dialysis treatment at home, during the year of follow-up. We found that individuals living in nonurban settings were more likely to transition to home modalities, with suburban communities having the highest home dialysis use.2 This may be related to known transportation challenges in suburban communities that rely on private vehicles and have limited public transportation options,15 or associations with income, education level, or having sufficient space to store dialysis supplies.
Our study findings are relevant to several key policies impacting dialysis care in the United States. First, our findings suggest that transportation is a clinically important domain that may require standardization of screening and improved support for people receiving dialysis. In previous studies of this cohort, we demonstrated that transportation mode was associated with missed dialysis appointments and higher mortality.16 In this analysis, a patient’s lack of private transportation was associated with a lower likelihood of transition to home dialysis. Our study highlights the importance of further understanding transportation insecurity among this high-risk population, as transportation impacts patients’ health outcomes and may influence dialysis modality choice.
Second, the Centers for Medicare and Medicaid Services has placed policy emphasis on increasing home dialysis use in the United States. Yet to date, financial incentives, such as the ETC model, have demonstrated little gain among Medicare beneficiaries. Our research suggests that this may be due in part to the role that social risks, such as transportation, play in modality choice. Importantly, the ETC model did not integrate support for addressing health-related social needs. Encouraging robust social risk screening and intervention may be a needed link to elevate home dialysis uptake. Addressing these social risks will be important not only among incident dialysis patients but also broadly among the entire dialysis population.
Strengths of the study include a large ESKD cohort with transportation information collected as part of routine care that includes individual-level transportation mode for patients treated with in-center HD. Yet our findings should be interpreted within the context of several limitations. First, individuals’ transportation mode may change over time and influence dialysis modality access. The use of only 1 transportation assessment at study entry may have influenced our results, although the impact was presumably limited, given that the follow-up time was restricted to a maximum of 1 year. Second, the study cohort consisted of adults who had already initiated in-center HD and been treated for a minimum of 90 days. As demonstrated by the time since the dialysis initiation variable, our cohort reflected a prevalent dialysis population, and the findings cannot be extrapolated to incident patients. According to the United States Renal Data System,2 among the prevalent dialysis population, the individuals most likely to be treated with home dialysis are Asian and White individuals, nondual Medicare and Medicaid beneficiaries, and those living in rural communities. Those individuals already treated with home dialysis at the study start were excluded from the cohort, and our study includes this selection bias. This is likely the reason why, in our cohort, despite Asian individuals having high home dialysis modality use nationally, they were less likely to transition to home dialysis. In addition, about 30% of patients treated with peritoneal dialysis and nearly 40% of patients treated with home HD convert to in-center HD within 2 years of home dialysis initiation, so our study likely included individuals with prior experience of home dialysis who may be unlikely to transition again to a home modality.2 Third, an individual’s insurance payer is associated with home dialysis utilization. Our research data set did not include payer type, and future studies characterizing the association between transportation and home dialysis utilization may benefit from including primary and secondary payer type. Fourth, lack of use of private transportation may be correlated with other social risks such as housing instability, financial needs, or utility difficulties, each of which may influence the likelihood of transitioning to home dialysis; therefore, our findings may relate to how social risk influences home dialysis use. Finally, we did not have information on caregiver support, which is a key consideration regarding modality choice.
Despite these limitations, our findings suggest important areas for further investigation. Upstream drivers of health, such as transportation, may influence patients’ access to health services across the kidney care spectrum from primary care to nephrology evaluation, home dialysis, and transplantation. Given growing interest surrounding screening and addressing social risks, such as transportation insecurity, dialysis care teams need further guidance on how to evaluate patients’ transportation needs and meaningful interventions that address these social needs. Our study highlights that individuals who are receiving in-center HD and who lack access to private transportation may have reduced access to home dialysis, even though this group may substantially benefit from home dialysis that could reduce the adverse consequences of missed dialysis related to transportation barriers. Advancing policies that provide flexible support to address social risks is an exciting direction to advance the use of home modalities. Home dialysis modalities are a potential intervention that provides individuals with more control, independence and reduces travel burden. Yet if individuals who lack private transportation are less likely to transition to home modalities, policy interventions that strive to advance home modality uptake without robust attention to upstream social risks may be limited in impact.
Article Information
Authors’ Full Names and Academic Degrees
Na’amah Razon, MD, PhD, Yi Zhang, PhD, Bethney Bonilla-Herrera, MA, Lorien S. Dalrymple, MD, MPH, Amanda K. Stennett, PhD, Baback Roshanravan, MD, MS, MSPH, Daniel Tancredi, PhD, and Joshua J. Fenton, MD, MPH
Authors’ Contributions
conceptualization: NR, LSD, BR, DT, JJF; data curation: NR, BBH, LSD, AKS, DT, JJF; formal analysis: NR, YZ, DT, JJF; funding acquisition: NR; investigation: NR; project administration: NR, BBH; resources: NR; validation: NR, YZ, BR, JJF; visualization: NR, YZ; methodology: NR, YZ, LSD, AKS, BR, DT, JJF; supervision: BR, DT, JJF. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.
Support
N. Razon and the project described were supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant number UL1TR001860 and linked award KL2 TR001859 and the National Institute of Diabetes and Digestion and Kidney Diseases K23 DK140602-01. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Financial Disclosure
L. Dalrymple is employed by and has shares/share-options in Fresenius Medical Care and owns stock in General Electric (GE) & GE HealthCare, and her spouse owns shares in The Permanente Medical Group. L. Dalrymple is a member of the Kidney Medicine Editorial Board. A. Stennett is employed by and has shares/share-options in Fresenius Medical Care. The remaining authors have no relevant financial interests to disclose.
Acknowledgments
The authors would like to acknowledge Laura Buckley and the EnRoute Community Advisory Board (Iris Corina, Bertha Dickerson, Maria Grijalva, Jovan Pulido, and the late Francine Williams [1955-2025]).
Data Sharing
The ability to share the study’s data is limited because of the existing data use agreement.
Peer Review
Received March 27, 2025. Evaluated by 2 external peer reviewers, with direct editorial input from an Associate Editor and the Editor-in-Chief. Accepted in revised form October 07, 2025.
Footnotes
Complete author and article information provided before references.
Table S1. STROBE Statement—Checklist of Items That Should Be Included in Reports of Cohort Studies.
Supplementary Material
Table S1.
References
- 1.Chan K., Thadhani R., Maddux F. Adherence barriers to chronic dialysis in the United States. J Am Soc Nephrol. 2014;25(11):2642–2648. doi: 10.1681/ASN.2013111160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.United States Renal Data System 2024 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. https://usrds-adr.niddk.nih.gov/2024
- 3.Au E., Davies C., Bateman S., et al. 47th Report, Chapter 2: Prevalence of Kidney Failure with Replacement Therapy. Australia and New Zealand Dialysis and Transplant Registry; 2024. https://anzdata.syd1.digitaloceanspaces.com/assets/common/02_prevalence_2023_ar_2024_chapter_F_20241206.pdf [Google Scholar]
- 4.Irish G., Davies C., Au E., et al. 47th Report, Chapter 1: Incidence of Kidney Failure with Replacement Therapy. Australia and New Zealand Dialysis and Transplant Registry. Australia and New Zealand Dialysis and Transplant Registry. https://anzdata.syd1.digitaloceanspaces.com/assets/common/01_incidence_2023_ar_2024_chapter_F4_20251125.pdf
- 5.US Department of Health and Human Services Advancing American Kidney Health. 2019. https://aspe.hhs.gov/sites/default/files/private/pdf/262046/AdvancingAmericanKidneyHealth.pdf
- 6.Centers for Medicare and Medicaid Services ESRD Treatment Choices (ETC) Model. https://www.cms.gov/priorities/innovation/innovation-models/esrd-treatment-choices-model
- 7.Johansen K.L., Li S., Liu J., et al. Association of the end-stage renal disease treatment choices payment model with home dialysis use at kidney failure onset from 2016 to 2022. JAMA Netw Open. 2023;6(2) doi: 10.1001/jamanetworkopen.2023.0806. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Negrusa B., Wiens J., Ullman D., et al. End-Stage Renal Disease Treatment Choices (ETC) Model- Second Annual Evaluation Report. The Lewin Group; Published online January 2024. Accessed 3 December 2024. https://www.cms.gov/priorities/innovation/data-and-reports/2024/etc-2nd-eval-rpt
- 9.Braveman P., Gottlieb L. The Social Determinants of Health: It’s Time to Consider the Causes of the Causes. Public Health Rep. 2014;129(Suppl 2):19–31. doi: 10.1177/00333549141291S206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.National Academies of Sciences, Engineering, and Medicine . The National Academies Press; 2019. Integrating Social Care into the Delivery of Health Care. [DOI] [PubMed] [Google Scholar]
- 11.Centers for Medicare and Medicaid Services Medicare Program; End-Stage Renal Disease Prospective Payment System, Payment for Renal Dialysis Services Furnished to Individuals With Acute Kidney Injury, Conditions for Coverage for End-Stage Renal Disease Facilities, End-Stage Renal Disease Quality Incentive Program, and End-Stage Renal Disease Treatment Choices Model. Published online November 12, 2024. https://www.govinfo.gov/content/pkg/FR-2024-11-12/pdf/2024-25486.pdf
- 12.Gottlieb L.M., DeSilvey S.C., Fichtenberg C., Bernheim S., Peltz A. Developing National Social Care Standards. Health Affairs Forefront. Published online February 22, 2023 doi: 10.1377/forefront.20230221.857308. [DOI] [Google Scholar]
- 13.U.S. Census Bureau American Community Survey Data 2016-2020. https://data.census.gov/table?q=educational+attainment
- 14.U.S Census Bureau. 2021 American Community Survey. https://data.census.gov/table?q=median+household+income
- 15.Sheller M. Racialized mobility transitions in Philadelphia: connecting urban sustainability and transport justice. City and Society. 2015;27(1):70–91. doi: 10.1111/ciso.12049. [DOI] [Google Scholar]
- 16.Razon N., Zhang Y., Bonilla-Herrera B., et al. Transportation insecurity and outcomes in hemodialysis patients: a retrospective cohort study. Clin J Am Soc Nephrol. 2025;20(8):1111–1118. doi: 10.2215/CJN.0000000753. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1.