Abstract
Rationale & Objective:
Beginning in 2014, some states expanded Medicaid eligibility to include additional low-income adults under age 65. Examining whether the expansion led to improvements in optimal treatment at the start of kidney failure has been understudied and was the focus of this investigation.
Study Design:
Observational cohort study.
Setting & Participants:
Patients aged 26–64 years represented in the United States Renal Data System initiating treatment for kidney failure between 2008 and 2019.
Exposure:
Residence in states that did or did not expand Medicaid in 2014.
Outcome:
Optimal start of kidney failure treatment was defined as undergoing preemptive kidney transplantation, initiating home dialysis, or initiating in-center hemodialysis using an arteriovenous access.
Analytical Approach:
Interrupted time series analyses were implemented to evaluate the adjusted association of residence in a Medicaid expansion state and optimal kidney failure treatment by comparing the trend in optimal kidney failure treatment starts during the pre-expansion vs. the post-expansion periods.
Results:
Before Medicaid expansion, the percentage with optimal starts increased similarly in expansion and non-expansion states. After Medicaid expansion, the percentage with optimal starts continued increasing in expansion states but decreased in non-expansion states, resulting in a 3.9% (95% CI 0.5%, 7.2%) higher percentage with optimal starts in expansion vs. non-expansion states by 2019 (P=0.02 for post vs. pre change in trend). Most of the change in trend was attributable to a greater increase in use of home dialysis at initiation of kidney failure treatment in expansion vs. non-expansion states (difference 0.29% per year [95% CI 0.08%, 0.51%]) during the post-expansion period.
Limitations:
Potential for unmeasured confounding from state-level factors other than Medicaid expansion.
Conclusions:
Medicaid expansion was associated with an increasing percentage of patients with incident kidney failure experiencing an optimal start to treatment, driven mostly by an increase in the use of home dialysis. Expanding Medicaid coverage may offer an opportunity to improve treatment for low-income patients initiating kidney replacement therapy.
Index words: arteriovenous access, home dialysis, in-center hemodialysis, kidney failure, Medicaid, optimal end stage renal disease starts, patient-centered care, preemptive kidney transplant, quality measures, uninsured
Plain-Language Summary
This study examined whether the expansion of Medicaid eligibility in 2014 led to improvements in the percentage of patients receiving an optimal start (i.e., kidney transplantation before the need to start dialysis, initiation of home dialysis, or initiation of in-center hemodialysis using an arteriovenous access) to treatment for new-onset kidney failure. We compared trends from 2008 to 2019 across U.S. states that expanded Medicaid and those that did not. Before 2014, both groups showed a similar upward trend in optimal starts. However, after Medicaid expansion, states that expanded Medicaid continued to experience a rise in optimal starts while states that did not expand showed a decline over time. Much of the difference was explained by a larger increase in the use of home dialysis in states that expanded access to Medicaid. Our findings suggest that expanding Medicaid may lead to better initial treatment for kidney failure.
Introduction
Prior to passage of the Patient Protection and Affordable Care Act (ACA),1 many adults under age 65 living in poverty were not eligible for Medicaid coverage.2 Although the ACA sought to extend eligibility to adults under age 65 with incomes up to 138% of the federal poverty level in all states, a 2012 U.S. Supreme Court decision effectively rendered expansion optional.3,4 Twenty-four states expanded Medicaid coverage under the ACA on January 1, 2014; 10 states had not done so as of November 2024.5–7
Many studies suggest that Medicaid expansion was associated with improvements in health-related outcomes including screening and access to care for chronic conditions such as diabetes and hypertension.8–13 Chronic kidney disease (CKD) affects about 14% of the adult population and a much higher percentage of those with diabetes mellitus and hypertension, the most common causes of CKD. Better care for individuals with diabetes, hypertension, and CKD, especially among those living in poverty,14–18 has the potential to slow progression of kidney disease to kidney failure and, importantly, to result in treatment of kidney failure that is more person-centered and better aligned with U.S. Government priorities.19
To contain costs and position patients to experience improved outcomes, new voluntary payment models introduced by the Centers for Medicare and Medicaid Services (CMS) incentivize an “optimal start” to treatment of kidney failure.20,21 As defined by CMS, the optimal start metric is the percentage of individuals who receive a preemptive kidney transplant, initiate dialysis using a home-based modality, or initiate in-center hemodialysis with an arteriovenous fistula or graft. Patients with CKD who progress to kidney failure without seeing a nephrologist are unlikely to have an optimal start to treatment. This is clear from vascular access rates: patients with over a year of nephrology care are four times as likely to begin hemodialysis with a permanent access as those with no or unknown duration of care (29% vs. 6.7%).22 We investigated whether the percentage of patients experiencing an optimal start to treatment of kidney failure increased in states that expanded Medicaid access in January 2014 compared with the percentage in states that did not expand Medicaid access before the end of 2019. We hypothesized that Medicaid expansion would be associated with a higher percentage of optimal starts. We further hypothesized that improvement in expansion states would be accompanied by better access to care from nephrologists and not driven by a higher overall rate of incident kidney failure.
Methods
Study Cohorts
We used data from the United States Renal Data System (USRDS),22 a national registry designed to track the incidence and prevalence of chronic kidney disease and treated kidney failure, to create yearly cohorts of patients with incident treated kidney failure from 2008 to 2019. The Hennepin Healthcare Human Research Protection Office approved this study as exempt secondary data research and waived the requirement for informed consent.
We identified patients primarily from the End Stage Renal Disease (ESRD) Medical Evidence Report (form CMS 2728), which providers are required to submit within 45 days of initiation of kidney replacement therapy. The form collects demographic and clinical characteristics at the time of kidney failure onset. We also used other USRDS data sources, including the patient event file in the ESRD Quality Reporting System (EQRS, formerly known as CROWNWeb) and the transplant file in the Organ Procurement and Transplantation Network (OPTN). We defined the date of kidney failure onset as the date of the first dialysis treatment for kidney failure or the date of receipt of a preemptive kidney transplant (i.e., a transplant prior to treatment with dialysis). We used Census Bureau data to estimate yearly counts of the U.S. population.
Classification of Medicaid Expansion Status
Individuals in each cohort were categorized as living in Medicaid “expansion” states or in “non-expansion” states. The former category included residents of the District of Columbia and 24 states that expanded Medicaid coverage on January 1, 2014, whereas the latter category included residents of 17 states that had not expanded Medicaid coverage by December 31, 2019. We excluded residents of 9 states that expanded Medicaid at an intermediate time during the study period (i.e., expansion after January 1, 2014, but before December 31, 2019) to create groups that were mutually exclusive throughout the study period. Lists of states in each category, as well as states that were excluded from the analysis, are presented in the footnote of Figure S1.
Defining Kidney Failure Outcomes
Incident kidney failure was defined as starting kidney replacement therapy (i.e., receiving dialysis or a kidney transplant). Patients who started kidney replacement therapy by receiving a preemptive kidney transplant, initiating home dialysis, or initiating in-center hemodialysis using an arteriovenous access were considered to experience an optimal start to treatment of kidney failure.23 For a small percentage (≤2.8%) of patients who initiated in-center hemodialysis, we were unable to determine whether their kidney failure treatment was optimal because they did not have a form CMS 2728. We also used form CMS 2728 to identify other characteristics at the start of kidney failure treatment, including having Medicaid coverage, being uninsured, and having received care from a nephrologist for ≥6 months before the onset of kidney failure.
Assessment of Covariates
Demographic variables, including age, sex, and combined categories of race and ethnicity, were defined using information from form CMS 2728 or other data sources. All other variables were ascertained from form CMS 2728. Clinical variables included the primary cause of kidney failure, body mass index, and selected comorbid conditions (diabetes, heart failure, atherosclerotic heart disease, cerebrovascular disease, peripheral vascular disease, and chronic obstructive pulmonary disease). We also included three measures of limitation in mobility or self-care, including (i) inability to ambulate or transfer, (ii) need for assistance with daily activities, and (iii) living in a nursing home or other institutionalized setting.
Statistical Analysis
Interrupted Time Series Analysis of the Percentage of Patients with Optimal Kidney Failure Treatment Starts and Related Outcomes
To evaluate the association between Medicaid expansion and change in trend of the percentage of incident patients who had optimal kidney failure treatment starts and related outcomes, we performed an interrupted time series analysis using yearly kidney failure cohorts. We excluded individuals aged <26 or ≥65 years as these persons were generally not the target of Medicaid expansion (Figure S1). In addition, to allow for ascertainment of clinical covariates, we excluded patients who did not have a form CMS 2728 within 90 days of the date of kidney failure onset. We also excluded patients who did not have a known modality of kidney replacement therapy.
We modeled optimal kidney failure treatment start as a binary outcome using a generalized linear model with a Gaussian distribution and identity link and using cluster-robust standard errors to account for within-state correlations.24 We modeled time as a continuous variable using a piecewise linear function with two pieces: the first from 2008 to 2013, and the second from 2014 to 2019. We fit an unadjusted model, which included main effects of each year (as a categorical variable) and expansion status and two-way interactions of expansion status with the two pieces of the time variable. Next, we fit models that were adjusted for demographic, clinical, and functional limitation covariates. We modeled age in 3 groups (26–44, 45–54, and 55–64 years). Analogously, we fit separate models for the following outcomes: receiving a preemptive kidney transplant, starting home dialysis, and starting in-center hemodialysis with a fistula or graft, as well as having Medicaid coverage, being uninsured, and having ≥6 months of nephrology care prior to the start of kidney failure treatment.
We also ran two sensitivity analyses to account for possible lag in policy impact and confounding (e.g., due to shortage of peritoneal dialysis fluid in 2014–2015).25 In these models, we treated 2014 and 2014–2015, respectively, as transition periods, excluding them from the pre- and post-expansion piecewise time variables.
Analysis of Optimal and Total Treated Kidney Failure Incidence
Trends in the percentage of patients with optimal kidney failure treatment starts could be influenced by the underlying incidence of kidney failure in the general population. We used kidney failure cohorts and data from the U.S. Census to evaluate the yearly incidence of total and optimal kidney failure treatment starts, as well as each component of optimal start for residents of Medicaid expansion and non-expansion states. For this analysis, we excluded only individuals aged <26 or ≥65 years.
We standardized incidence estimates to the demographic characteristics of the 2014 Census population using a model-based method.26 As in the first analysis, we fit generalized linear models with a Gaussian distribution and identity link and with cluster-robust standard errors to account for state-level variation. Each model used the yearly kidney failure incidence rate as the outcome and was weighted by the population size of the respective demographic subgroup. The models included main effects of year, expansion status, age, sex, and race or ethnicity; the interaction of year and expansion status; all two-way interactions involving year with age, sex, and race or ethnicity; all two-way interactions involving expansion status with age, sex, and race or ethnicity; and all three-way interactions involving year and expansion status with age, sex, and race or ethnicity.
Results
Patient Characteristics at the Start of Kidney Failure Treatment
Construction of yearly cohorts of incident patients treated for kidney failure used to assess the association of Medicaid expansion with the percentage of optimal kidney failure treatment starts is shown in Figure S1. In total, we included 323,807 patients from expansion states (155,373 pre-expansion; 168,434 post-expansion) and 271,725 patients from non-expansion states (128,842 pre-expansion; 142,883 post-expansion). The percentage of non-Hispanic Black individuals was lower and the percentage of individuals of non-Hispanic White, Hispanic, and other race or ethnicity groups were higher in expansion states (Table 1). Hypertension as the primary cause of kidney failure, obesity, and cerebrovascular disease were less common among patients in expansion states, whereas living in an institutionalized setting was more common.
Table 1.
Characteristics of Patients With Incident Kidney Failure in Selected Years, by Residence in States That Did or Did Not Expand Medicaid.
| Pre-expansion period | Post-expansion period | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2008 | 2011 | 2013 | 2014 | 2017 | 2019 | |||||||
| Medicaid expansion | No | Yes | No | Yes | No | Yes | No | Yes | No | Yes | No | Yes |
| Number of patients | 20,851 | 25,210 | 21,147 | 25,887 | 22,361 | 26,188 | 22,665 | 27,135 | 23,802 | 27,947 | 24,775 | 28,863 |
| Age (years) | ||||||||||||
| Mean | 51.8 | 52.2 | 52.3 | 52.5 | 52.3 | 52.5 | 52.1 | 52.5 | 52.3 | 52.5 | 52.1 | 52.5 |
| Standard deviation | 9.5 | 9.5 | 9.5 | 9.5 | 9.5 | 9.5 | 9.6 | 9.5 | 9.6 | 9.6 | 9.7 | 9.7 |
| Age category (years), % | ||||||||||||
| 26–44 | 23.3 | 22.0 | 22.4 | 20.9 | 22.1 | 20.8 | 22.7 | 21.3 | 22.1 | 21.2 | 23.0 | 21.5 |
| 45–54 | 31.5 | 30.9 | 30.1 | 31.1 | 30.2 | 30.8 | 30.3 | 29.5 | 29.9 | 29.4 | 29.7 | 28.7 |
| 55–64 | 45.1 | 47.2 | 47.5 | 48.0 | 47.7 | 48.4 | 47.1 | 49.1 | 48.0 | 49.4 | 47.2 | 49.7 |
| Female, % | 43.5 | 41.1 | 42.1 | 39.3 | 41.9 | 39.5 | 41.9 | 39.0 | 41.1 | 38.5 | 41.5 | 38.1 |
| Race or ethnicity, % | ||||||||||||
| Non-Hispanic, White | 39.3 | 44.3 | 39.0 | 42.9 | 39.8 | 43.9 | 40.6 | 43.3 | 42.1 | 43.8 | 40.3 | 41.5 |
| Non-Hispanic, Black | 42.7 | 27.3 | 41.2 | 26.3 | 40.7 | 24.8 | 39.6 | 24.9 | 37.4 | 22.6 | 37.1 | 22.3 |
| Hispanic | 15.0 | 19.3 | 16.3 | 21.3 | 16.2 | 21.5 | 16.4 | 21.7 | 17.1 | 22.6 | 18.8 | 24.7 |
| Non-Hispanic, Other | 3.1 | 9.1 | 3.4 | 9.5 | 3.2 | 9.9 | 3.4 | 10.0 | 3.5 | 11.0 | 3.9 | 11.6 |
| Primary cause of kidney failure, % | ||||||||||||
| Diabetes | 48.3 | 48.5 | 48.6 | 48.8 | 49.1 | 49.5 | 48.8 | 49.5 | 50.0 | 51.6 | 49.8 | 50.7 |
| Hypertension | 24.7 | 19.6 | 26.1 | 20.6 | 28.7 | 23.2 | 28.8 | 23.5 | 29.4 | 23.3 | 28.9 | 22.8 |
| Glomerulonephritis | 10.0 | 11.7 | 9.6 | 11.5 | 8.7 | 10.3 | 8.4 | 10.1 | 7.9 | 9.9 | 7.5 | 9.4 |
| Cystic kidney | 3.5 | 3.9 | 3.2 | 3.4 | 2.9 | 3.5 | 2.9 | 3.6 | 3.5 | 4.1 | 3.4 | 4.0 |
| Other/unknown | 13.6 | 16.2 | 12.5 | 15.7 | 10.7 | 13.4 | 11.1 | 13.3 | 9.2 | 11.2 | 10.4 | 13.0 |
| Body mass index (kg/m2), % | ||||||||||||
| <18.5 | 2.8 | 3.3 | 2.6 | 2.6 | 2.7 | 2.5 | 2.4 | 2.4 | 2.4 | 2.3 | 2.2 | 2.5 |
| 18.5 to <25 | 24.6 | 27.6 | 22.8 | 26.2 | 23.1 | 25.7 | 22.9 | 26.3 | 22.7 | 25.4 | 22.1 | 24.9 |
| 25 to <30 | 26.4 | 27.0 | 26.5 | 27.0 | 25.8 | 27.4 | 26.5 | 27.4 | 26.1 | 27.4 | 26.0 | 27.9 |
| 30 to <35 | 19.7 | 18.6 | 20.5 | 19.6 | 20.7 | 19.7 | 20.6 | 19.7 | 20.9 | 20.6 | 21.7 | 20.3 |
| 35 to <40 | 11.9 | 10.9 | 13.0 | 11.4 | 12.6 | 11.8 | 13.1 | 11.8 | 13.3 | 11.7 | 13.3 | 12.0 |
| ≥40 | 13.2 | 11.4 | 13.6 | 12.2 | 14.4 | 11.9 | 13.8 | 11.8 | 14.2 | 12.1 | 14.3 | 12.1 |
| Unknown | 1.4 | 1.1 | 1.0 | 1.0 | 0.8 | 0.9 | 0.7 | 0.6 | 0.5 | 0.5 | 0.4 | 0.4 |
| Comorbid conditions, % | ||||||||||||
| Diabetes | 56.2 | 53.9 | 58.4 | 56.3 | 59.4 | 57.3 | 59.2 | 57.2 | 61.0 | 58.7 | 61.8 | 60.1 |
| Heart failure | 24.6 | 22.9 | 24.2 | 22.1 | 22.2 | 20.6 | 22.1 | 20.2 | 22.3 | 20.1 | 24.0 | 21.3 |
| Atherosclerotic heart disease | 12.9 | 13.1 | 12.0 | 12.0 | 9.1 | 10.0 | 8.5 | 8.7 | 8.0 | 8.1 | 7.6 | 7.9 |
| Cerebrovascular disease | 8.1 | 6.4 | 7.7 | 6.4 | 7.4 | 6.0 | 7.4 | 5.8 | 7.9 | 6.2 | 7.7 | 6.8 |
| Peripheral vascular disease | 10.3 | 9.8 | 10.1 | 8.8 | 8.7 | 7.5 | 8.0 | 7.5 | 7.2 | 7.0 | 7.5 | 6.9 |
| Chronic obstructive pulmonary disease | 5.9 | 5.7 | 6.1 | 5.8 | 6.3 | 5.8 | 6.4 | 5.1 | 6.1 | 5.3 | 6.1 | 5.5 |
| Functional limitations, % | ||||||||||||
| Inability to ambulate or transfer | 4.9 | 4.4 | 4.9 | 4.8 | 4.8 | 4.3 | 4.7 | 4.4 | 5.1 | 4.9 | 5.3 | 4.8 |
| Needs assistance with daily activities | 7.6 | 7.3 | 8.4 | 8.6 | 8.8 | 8.7 | 8.3 | 8.3 | 9.4 | 9.5 | 9.7 | 9.6 |
| Lives in institutionalized setting | 4.0 | 4.5 | 4.3 | 5.3 | 4.0 | 5.1 | 4.1 | 4.7 | 4.2 | 5.0 | 3.7 | 4.3 |
Association between Medicaid Expansion and the Percentage of Optimal Kidney Failure Treatment Starts
Analysis of unadjusted trends in the percentage of optimal kidney failure treatment starts demonstrates sound model fit based on visual inspection (Figure S2). In the pre-Medicaid expansion period (2008 to 2013), the adjusted percentage of optimal starts increased in both expansion and non-expansion states (Figure 1a), and there was no difference in the trend (Table 2). After Medicaid expansion, adjusted percentage of optimal starts in expansion states continued rising, whereas they decreased in non-expansion states (adjusted percentage difference: 3.9% [95% CI 0.5%, 7.2%] in 2019) (Figure 1a, Table 2). This difference in trend in the post-expansion period was statistically significant (0.55% [95% CI 0.14%, 0.96%] per year higher adjusted percentage difference in expansion vs. non-expansion states [post vs. pre change in trend: P=0.02]) (Table 2).
Figure 1.
Fully-adjusted yearly percentage of optimal kidney failure treatment starts, overall and by component, by residence in states that did or did not expand Medicaid. The circles and lines together represent predictions from generalized linear regression models that are adjusted for age, sex, race or ethnicity, primary cause of kidney failure, body mass index, comorbid conditions, and functional limitations. Models use cluster-robust standard errors at the state level. The shaded areas represent 95% confidence intervals.
Table 2.
Fully-Adjusted Yearly Percentage Differences of Optimal Kidney Failure Treatment Starts, Overall and by Component, by Residence in States That Did or Did Not Expand Medicaid.
| (a) Optimal kidney failure treatment start | (b) Kidney transplant | (c) Home dialysis | (d) In-center HD with arteriovenous access | |||||
|---|---|---|---|---|---|---|---|---|
| Est. | (95% CI) | Est. | (95% CI) | Est. | (95% CI) | Est. | (95% CI) | |
| Percentage difference* (expansion vs. non-expansion [ref]) | ||||||||
| 2008 | 0.5 | (−1.8, 2.8) | 0.4 | (−0.5, 1.3) | −1.1 | (−2.3, −0.0) | 1.2 | (−0.7, 3.1) |
| 2009 | 0.5 | (−1.7, 2.8) | 0.4 | (−0.4, 1.2) | −1.3 | (−2.5, −0.1) | 1.4 | (−0.5, 3.3) |
| 2010 | 0.5 | (−1.7, 2.8) | 0.4 | (−0.3, 1.2) | −1.5 | (−2.7, −0.2) | 1.6 | (−0.4, 3.6) |
| 2011 | 0.5 | (−1.7, 2.8) | 0.4 | (−0.3, 1.1) | −1.6 | (−3.0, −0.2) | 1.7 | (−0.4, 3.9) |
| 2012 | 0.6 | (−1.8, 2.9) | 0.4 | (−0.3, 1.1) | −1.8 | (−3.4, −0.2) | 1.9 | (−0.3, 4.2) |
| 2013 | 0.6 | (−1.9, 3.1) | 0.4 | (−0.2, 1.0) | −1.9 | (−3.7, −0.1) | 2.1 | (−0.3, 4.5) |
| 2014 | 1.1 | (−1.4, 3.6) | 0.4 | (−0.2, 1.0) | −1.6 | (−3.5, 0.2) | 2.3 | (0.0, 4.7) |
| 2015 | 1.7 | (−0.9, 4.2) | 0.4 | (−0.2, 1.0) | −1.3 | (−3.3, 0.6) | 2.6 | (0.4, 4.8) |
| 2016 | 2.2 | (−0.5, 4.9) | 0.4 | (−0.2, 1.1) | −1.0 | (−3.2, 1.1) | 2.8 | (0.7, 5.0) |
| 2017 | 2.8 | (−0.1, 5.6) | 0.5 | (−0.2, 1.1) | −0.8 | (−3.0, 1.5) | 3.1 | (0.9, 5.2) |
| 2018 | 3.3 | (0.2, 6.4) | 0.5 | (−0.3, 1.2) | −0.5 | (−2.9, 1.9) | 3.3 | (1.2, 5.4) |
| 2019 | 3.9 | (0.5, 7.2) | 0.5 | (−0.3, 1.3) | −0.2 | (−2.7, 2.4) | 3.5 | (1.4, 5.7) |
| Trend in percentage difference † | ||||||||
| Pre-expansion trend, per year | 0.01 | (−0.32, 0.34) | −0.01 | (−0.12, 0.10) | −0.16 | (−0.44, −0.13) | 0.18 | (−0.11, 0.46) |
| Post-expansion trend, per year | 0.55 | (0.14, 0.96) | 0.02 | (−0.11, 0.14) | 0.29 | (0.08, 0.51) | 0.24 | (−0.04, 0.52) |
| Change in trend, per year | 0.53 | (0.07, 1.00) | 0.02 | (−0.18, 0.23) | 0.45 | (0.13, 0.77) | 0.06 | (−0.40, 0.52) |
| P-value for change in trend | P=0.02 | P=0.86 | P=0.007 | P=0.87 | ||||
Estimates are adjusted for age, sex, race or ethnicity, primary cause of kidney failure, body mass index, comorbid conditions, and functional limitations using generalized linear regression models with cluster-robust standard errors at the state level.
The percentage difference estimates in the upper part of the table are yearly adjusted model predictions.
The trend in percentage difference estimates in the lower part of the table are based on the model coefficients for the interaction of each piecewise time term with Medicaid expansion status.
Most of the change in trend in overall optimal starts was attributable to trends in home dialysis starts (Figure 1, Table 2). In the pre-expansion period, the adjusted percentage of patients initiating home dialysis was lower in expansion states than in non-expansion states and was increasing to a slightly lesser extent (Figure 1c, Table 2). In the post-expansion period, the trend in home dialysis starts increased more rapidly in expansion states than in non-expansion states (difference 0.29% per year [95% CI 0.08%, 0.51%], P=0.007, Table 2), such that by 2019, the adjusted percentage was nearly equivalent in patients in expansion (13.8%) and non-expansion (14.0%) states (Figure 1c).
Across the entire study period, a higher percentage of patients in expansion states received a preemptive kidney transplant and started in-center hemodialysis with an arteriovenous access than in non-expansion states (Figure 1b, Figure 1d). The percentage of patients starting in-center hemodialysis with an arteriovenous access was increasing in expansion states relative to non-expansion states throughout the entire study period, but there was no evidence of a change in trend for these outcomes related to Medicaid expansion (post vs. pre change in trends P=0.86 for kidney transplant and P=0.87 for use of an arteriovenous access for in-center hemodialysis) (Table 2). There was no association of Medicaid expansion with preemptive kidney transplant (Figure 1b, Table 2).
The association between Medicaid expansion and optimal starts remained consistent in two sensitivity analyses in which the years 2014 and 2014–2015, respectively, were treated as transition years (Table S1).
Association between Medicaid Expansion and Trends in Insurance Coverage and Nephrology Care
In the pre-expansion period, a higher percentage of patients in expansion states had Medicaid coverage at the start of kidney failure treatment (33.3% to 37.5%) than in non-expansion states (25.5% to 26.2%) in the adjusted analysis (Figure 2a; see Figure S3 for unadjusted results). After Medicaid expansion, Medicaid coverage increased in patients residing in expansion states, reaching 46.7% by 2019, whereas it fell to 23.2% in patients in non-expansion states (adjusted percentage difference: 23.5% [95% CI 16.0%, 30.9%]) (Figure 2a, Table S2). The adjusted percentage of patients with Medicaid coverage increased by 1.97% [95% CI 1.60, 2.34%] more per year in the expansion states in the post-expansion period (post vs. pre change in trend: P<0.001) (Table S2). Conversely, the adjusted percentage of patients who were uninsured at the start of kidney failure treatment decreased by 1.04% (95% CI 0.69%, 1.40%) more per year in expansion states than in non-expansion states after Medicaid expansion (post vs pre change in trend: P<0.001) (Figure 2b, Table S2). The adjusted percentage of patients who had ≥6 months of nephrology care prior to the onset of kidney failure in expansion states increased more than in non-expansion states during the study period (Figure 2c). However, the change in trend actually decreased slightly after Medicaid expansion (Table S2).
Figure 2.
Fully-adjusted yearly percentages of kidney failure treatment starts with Medicaid, without insurance, and with prior nephrology care, by residence in states that did or did not expand Medicaid. The circles and lines together represent predictions from generalized linear regression models that are adjusted for age, sex, race or ethnicity, primary cause of kidney failure, body mass index, comorbid conditions, and functional limitations. Models use cluster-robust standard errors at the state level. The shaded areas represent 95% confidence intervals.
Incidence of Kidney Failure Treatment
Construction of the yearly cohorts of incident patients with kidney failure and the corresponding U.S. Census populations used for analyses of incident kidney failure is shown in Figure S4. Prior to Medicaid expansion, the incidence of optimal kidney failure treatment starts was lower in expansion states (92 to 104 cases per million) than in non-expansion states (98 to 115 cases per million) and showed a pattern of increase in both groups (Figure S5a). After expansion, the incidence of optimal kidney failure treatment starts continued to increase in expansion states but stabilized in non-expansion states, leading to incidence estimates that were nearly overlapping from 2017–2019 (Figure S5a). Across the study period, the incidence of total kidney failure treatment starts varied, but point estimates were consistently lower in states that expanded Medicaid (ranging from 344 to 364 cases per million) than in states that did not (374 to 409 cases per million) (Figure S5b). Patterns of incidence of each of the components of optimal kidney failure treatment (Figure S6) mirrored those of the percentage of those outcomes (Figure S2) except that population-level rates of starting in-center hemodialysis using an arteriovenous access diverged after 2014. There was virtually no difference between expansion and non-expansion states in this outcome from 2008–2014, but after 2014 point estimates were consistently higher in expansion states (Figure S6c).
Discussion
The percentage of patients with incident kidney failure experiencing an optimal start to treatment improved after 2014 in states that expanded Medicaid access relative to those that did not. Prior to expansion, the percentage of patients experiencing an optimal start was similar in expansion and non-expansion states and was rising. After 2014, the percentage continued to increase in expansion states while declining in non-expansion states (Figure 1). The divergence in the trajectory of optimal starts after Medicaid expansion was due mainly to greater increases in use of home dialysis and continuation of the faster rate of uptake of arteriovenous access among patients starting in-center hemodialysis in expansion states. The percentage of patients receiving a preemptive kidney transplant was higher in expansion states throughout the study period, but the difference between expansion and non-expansion states did not change after Medicaid expansion.
Improvements in optimal starts to treatment of kidney failure in Medicaid expansion states were accompanied by expected increases in the percentage of patients covered by Medicaid (and reciprocal decreases in the percentage uninsured), similar to what was reported by Swaminathan et al through 2017.27 The percentage with nephrology care increased in expansion states relative to non-expansion states throughout the study period, but the trend did not intensify after 2014. Nevertheless, the percentage receiving nephrology care in expansion states exceeded the percentage in non-expansion states by 2016. Greater access to nephrology care likely contributed to the improvement in optimal starts in expansion states. However, because access to nephrology care was higher in non-expansion than expansion states prior to 2014, better access to nephrology care after expansion was likely not the only important factor. Other possible contributors include higher intensity or frequency of nephrology care, more education about treatment of kidney failure, and access to dialysis access procedures, among others.
In theory, an increase in the percentage of optimal starts to treatment of kidney failure in expansion states could be the result of growth in overall rates of treatment for kidney failure – which would not be a positive outcome even if the increase were mostly in patients receiving optimal treatment. However, our examination of rates of kidney failure did not support this possibility; the rate of optimal treatment increased in expansion states relative to non-expansion states without an increase in the overall rate of kidney failure treatment (Figure S5).
Although optimal start to treatment of kidney failure increased in expansion states relative to non-expansion states after 2014, the percentage of patients treated with preemptive kidney transplant and in-center hemodialysis with a permanent access were already higher in expansion states before Medicaid expansion. One potential explanation is that the percentage uninsured was already lower in expansion states before 2014 for individuals with kidney disease (Figure 2b) and for all non-elderly adults with incomes below 200% of the poverty line (35% in 2013 versus 44% in states that did not expand Medicaid access before 2022).28
It is imperative to understand the relationship between socioeconomic factors, such as insurance status, and outcomes in people with kidney disease.14 Previous work has shown that states with broader coverage with Medicaid have lower kidney failure incidence29 and that insurance with Medicaid (versus no insurance) is associated with more rapid placement of an arteriovenous access after hemodialysis initiation.30 Is it therefore reasonable to suggest that Medicaid expansion might impact kidney disease-related outcomes. Indeed, Harhay et al reported that Medicaid expansion was associated with increases in preemptive kidney transplantation.31 Thorsness et al found a decrease in incident kidney failure in expansion states relative to non-expansion states in the early post-expansion period that was not sustained, but they did not examine patterns of treatment of kidney failure.32 Another group reported fewer tunneled dialysis catheters (and therefore more arteriovenous accesses) among patients starting hemodialysis in expansion states than non-expansion states in the post expansion period.33 However, rates of catheter use were higher in non-expansion states even before 2014. Analogously, Swaminathan et al found that initiation of hemodialysis with presence of an arteriovenous access decreased in non-expansion states relative to expansion states after 2014.27 Thus, our study adds important context necessary to understand whether outcomes improved in states that expanded Medicaid.
Our study has limitations. We stopped follow-up at the end of 2019 because the COVID-19 pandemic limited access to kidney transplant and arteriovenous fistula placement in 2020,34,35 and differences in rates of COVID-19 infections and hospitalizations in each state would likely be major drivers of outcomes. Therefore, we elected to consider the period prior to this catastrophically disruptive event. We did not examine outcomes after the start of treatment of kidney failure. However, optimal start to treatment of kidney failure is itself a positive outcome for patients,36 and our findings may partially explain better outcomes reported in the first year of treatment for kidney failure.27 We also cannot rule out the possibility that some of our findings are explained by other interventions in expansion states beyond that of Medicaid expansion itself. Interventions related to housing, transportation, or education, which likely impact social determinants of health, could, in turn, be associated with optimal starts to treatment of kidney failure. However, we are not aware of systematic initiatives of this sort across expansion states.
In conclusion, we found that the percentage of patients with incident kidney failure receiving a pre-emptive kidney transplant, home dialysis, or in-center hemodialysis with an arteriovenous access increased after 2014 in states that expanded Medicaid access relative to those that did not. The difference in the trajectory of optimal treatment from pre- to post-expansion was driven mainly by an increase in the percentage using home dialysis in expansion states after 2014. The rate of increase in use of permanent vascular access in expansion states outpaced that of non-expansion states both before and after 2014. These differences in treatment were not driven by differences in overall rates of kidney failure.
Supplementary Material
Figure S1. Cohort construction for analyses of the percentage of patients with optimal kidney failure treatment starts and related outcomes.
Figure S2. Unadjusted yearly percentages of optimal kidney failure treatment starts, overall and by component, by residence in states that did or did not expand Medicaid.
Figure S3. Unadjusted yearly percentages of kidney failure treatment starts with Medicaid, without insurance, and with prior nephrology care, by residence in states that did or did not expand Medicaid.
Figure S4. Cohort construction for analyses of the incidence of total and optimal kidney failure treatment starts in the general US population.
Figure S5. Adjusted yearly incidence of optimal and total kidney failure treatment starts by residence in states that did or did not expand Medicaid.
Figure S6. Adjusted yearly incidence of each optimal kidney failure treatment start component by residence in states that did or did not expand Medicaid.
Table S1. Fully-adjusted trends in percentage difference of optimal kidney failure treatment starts, overall and by component, by residence in states that did or did not expand Medicaid (sensitivity analysis treating 2014 and 2014–2015 as transition years).
Table S2. Fully-adjusted yearly percentage differences of kidney failure treatment starts with Medicaid, without insurance, and with prior nephrology care, by residence in states that did or did not expand Medicaid.
Acknowledgments:
The data reported here have been supplied by the Centers for Medicare & Medicaid Services (CMS) to the USRDS under a data use agreement between CMS and NIDDK. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the U.S. government.
Support:
This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (contract no. N01 75N94019C00006 and N01 75N94024C00001). The funders had no role in study design, data collection, data analysis, interpretation and reporting of results, or in decisions to submit for publication.
Financial Disclosure:
Dr Roetker receives grant/research support from Fresenius, Merck, and the National Institutes of Health (NIDDK and National Heart, Lung, and Blood Institute [NHLBI]). Dr Wetmore receives grant/research support from NIDDK, NHLBI, GSK, Merck, AstraZeneca, and Fresenius. Dr Liu receives grant/research support from GSK, Amgen, and the National Institutes of Health. Ms Guo has received grant/research support from Gilead, Amgen, and GSK. Dr Gilbertson receives grant/research support from Fresenius, Gilead, GSK, Amgen, and NIDDK. Dr Johansen receives grant/research support from NIDDK and consults with GSK and Akebia.
Footnotes
Declaration of Generative AI and AI-Assisted Technologies in the Writing Process: During the preparation of this work the authors used Gemini 2.5 Pro, accessed via gemini.google.com, in order to minimize word count. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.
Peer Review: Received June 3, 2025. Evaluated by 3 external peer reviewers, with direct editorial input from a Statistics/Methods Editor, an Associate Editor, and the Editor-in-Chief. Accepted in revised form January 26, 2026.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Sharing:
The individual patient-level data underlying the results presented in this manuscript are freely available, subject to terms of a data use agreement with the United States Renal Data System and NIDDK, to qualified individuals or institutions.
References
- 1.111th Congress. The Patient Protection and Affordable Care Act. In: The 111th U.S. Congress, ed. Public Law 111–148. Vol 42 USC 18001, 2010. [Google Scholar]
- 2.Healthinsurance.org. Medicaid expansion. Accessed January 24, 2025, https://www.healthinsurance.org/glossary/medicaid-expansion/.
- 3.Rosenbaum S, Westmoreland TM. The Supreme Court’s Surprising Decision On The Medicaid Expansion: How Will The Federal Government And States Proceed? Health Affairs. Aug 2012;31(8):1663–1672. doi: 10.1377/hlthaff.2012.0766 [DOI] [PubMed] [Google Scholar]
- 4.National Federation of Independent Business v. Sebelius, 567 U.S. 519, (U.S. Supreme Court 2012). https://supreme.justia.com/cases/federal/us/567/519/ [Google Scholar]
- 5.Congressional Research Service. Overview of the ACA Medicaid expansion. crsresports.congress.gov: Congressional Research Service; 2021. https://crsreports.congress.gov/product/pdf/IF/IF10399 [Google Scholar]
- 6.KFF. Status of state Medicaid expansion decisions: Interactive map. Accessed April 7, 2024, https://www.kff.org/affordable-care-act/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map/
- 7.Tolbert J, Bell C, Rudowitz R. Medicaid expansion is a red and blue state issue. KFF. Accessed February 2, 2025, https://www.kff.org/medicaid/issue-brief/medicaid-expansion-is-a-red-and-blue-state-issue/
- 8.Guth M, Ammula M. Building on the evidence base: Studies on the effects of Medicaid expansion, February 2020 to March 2021. KFF. Accessed 4/7/2024, https://www.kff.org/medicaid/report/building-on-the-evidence-base-studies-on-the-effects-of-medicaid-expansion-february-2020-to-march-2021/
- 9.Guth M, Garfield R, Rudowitz R. The effects of Medicaid expansion under the ACA: Studies from January 2014 to January 2020. KFF. Accessed January 25, 2025, https://www.kff.org/medicaid/report/the-effects-of-medicaid-expansion-under-the-aca-updated-findings-from-a-literature-review/
- 10.Yue D, Zhu Y, Rasmussen PW, Godwin J, Ponce NA. Coverage, Affordability, and Care for Low-Income People with Diabetes: 4 Years after the Affordable Care Act’s Medicaid Expansions. Journal of General Internal Medicine. Jul 2020;35(7):2222–2224. doi: 10.1007/s11606-019-05614-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yan LD, Ali MK, Strombotne KL. Impact of Expanded Medicaid Eligibility on the Diabetes Continuum of Care Among Low-Income Adults: A Difference-in-Differences Analysis. American Journal of Preventive Medicine. Feb 2021;60(2):189–197. doi: 10.1016/j.amepre.2020.08.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sumarsono A, Buckley LF, Machado SR, et al. Medicaid Expansion and Utilization of Antihyperglycemic Therapies. Diabetes Care. Nov 2020;43(11):2684–2690. doi: 10.2337/dc20-0735 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Heather A, Nathalie H, David E-H, et al. New hypertension and diabetes diagnoses following the Affordable Care Act Medicaid expansion. Family Medicine and Community Health. Dec 2020;8(4):e000607. doi: 10.1136/fmch-2020-000607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rizzolo K, Shen JI. Barriers to home dialysis and kidney transplantation for socially disadvantaged individuals. Curr Opin Nephrol Hypertens. Jan 1 2024;33(1):26–33. doi: 10.1097/MNH.0000000000000939 [DOI] [PubMed] [Google Scholar]
- 15.Patzer RE, McClellan WM. Influence of race, ethnicity and socioeconomic status on kidney disease. Nat Rev Nephrol. Sep 2012;8(9):533–41. doi: 10.1038/nrneph.2012.117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.McPherson L, Plantinga LC, Howards PP, Kramer M, Patzer RE. Effect of pre-end-stage kidney disease nephrology care on the association between neighborhood poverty and referral for kidney transplantation among patients with end-stage kidney disease. J Clin Epidemiol. Sep 2025;185:111878. doi: 10.1016/j.jclinepi.2025.111878 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Grobman B, Turkson-Ocran RA, Zhang M, Juraschek SP. Socioeconomic Status and Chronic Kidney Disease Among Black and White Adults: An Analysis of 2017–2020 NHANES. Kidney Med. Aug 2025;7(8):101045. doi: 10.1016/j.xkme.2025.101045 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Cervantes L, Sinclair M, Camacho C, Santana C, Novick T, Cukor D. Social and Behavioral Barriers to Effective Care During the Transition to End-Stage Kidney Care. Adv Kidney Dis Health. Jan 2024;31(1):21–27. doi: 10.1053/j.akdh.2023.09.005 [DOI] [PubMed] [Google Scholar]
- 19.U.S. Department of Health and Human Services. Advancing American Kidney Health. 2019. https://aspe.hhs.gov/system/files/pdf/262046/AdvancingAmericanKidneyHealth.pdf [Google Scholar]
- 20.National Quality Forum (NQF). NQP Playbook: Supporting Shared Decision Making for Individuals with Chronic Kidney Disease and End-Stage Renal Disease. 2019. https://www.qualityforum.org/
- 21.Jain G, Weiner DE. Value-Based Care in Nephrology: The Kidney Care Choices Model and Other Reforms. Kidney360. Oct 28 2021;2(10):1677–1683. doi: 10.34067/kid.0004552021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Johansen KL, Gilbertson DT, Li S, et al. US Renal Data System 2023 Annual Data Report: Epidemiology of Kidney Disease in the United States. Am J Kidney Dis. 2024;83(4S1):A8–A13. doi: 10.1053/j.ajkd.2024.01.001 [DOI] [PubMed] [Google Scholar]
- 23.Centers for Medicare & Medicaid Services. PY2023 Request for Applications (RFA). Updated March 6, 2022. Accessed January 9, 2025, https://www.cms.gov/priorities/innovation/media/document/kcc-py23-rfa
- 24.Naimi AI, Whitcomb BW. Estimating Risk Ratios and Risk Differences Using Regression. Am J Epidemiol. Jun 1 2020;189(6):508–510. doi: 10.1093/aje/kwaa044 [DOI] [PubMed] [Google Scholar]
- 25.Jensen V, Throckmorton DC. Shortage of Peritoneal Dialysis Solution and the Food and Drug Administration’s Response. Clin J Am Soc Nephrol. Aug 7 2015;10(8):1484–6. doi: 10.2215/CJN.12061214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Liu J, Louis TA, Pan W, Ma JZ, Collins AJ. State-level adjusted ESRD incident rates: use of observed vs model-predicted category-specific rates. Kidney Int. Apr 2006;69(8):1459–63. doi: 10.1038/sj.ki.5000299 [DOI] [PubMed] [Google Scholar]
- 27.Swaminathan S, Sommers BD, Thorsness R, Mehrotra R, Lee Y, Trivedi AN. Association of Medicaid Expansion With 1-Year Mortality Among Patients With End-Stage Renal Disease. JAMA. Dec 4 2018;320(21):2242–2250. doi: 10.1001/jama.2018.16504 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Center on Budget and Policy Priorities. Expansion states saw large drop in uninsured rates. Center on Budget and Policy Priorities. Accessed January 26, 2025, https://www.cbpp.org/charts/expansion-states-saw-large-drop-in-uninsured-rates-3
- 29.Kurella-Tamura M, Goldstein BA, Hall YN, Mitani AA, Winkelmayer WC. State medicaid coverage, ESRD incidence, and access to care. J Am Soc Nephrol. Jun 2014;25(6):1321–9. doi: 10.1681/ASN.2013060658 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Lin E, Mell MW, Winkelmayer WC, Erickson KF. Health Insurance in the First 3 Months of Hemodialysis and Early Vascular Access. Clin J Am Soc Nephrol. Dec 7 2018;13(12):1866–1875. doi: 10.2215/CJN.06660518 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Harhay MN, McKenna RM, Harhay MO. Association Between Medicaid Expansion Under the Affordable Care Act and Medicaid-Covered Pre-emptive Kidney Transplantation. J Gen Intern Med. Nov 2019;34(11):2322–2325. doi: 10.1007/s11606-019-05279-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Thorsness R, Swaminathan S, Lee Y, et al. Medicaid Expansion and Incidence of Kidney Failure among Nonelderly Adults. Journal of the American Society of Nephrology. Jun 1 2021;32(6):1425–1435. doi: 10.1681/asn.2020101511 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Levin SR, Farber A, Eslami MH, et al. Association of Medicaid Expansion with Tunneled Dialysis Catheter Use at the Time of First Arteriovenous Access Creation. Annals of Vascular Surgery. Jul 2021;74:11–20. doi: 10.1016/j.avsg.2021.01.063 [DOI] [PubMed] [Google Scholar]
- 34.Allon M, Zhang Y, Thamer M, Crews DC, Lee T. Trends in Vascular Access Among Patients Initiating Hemodialysis in the US. JAMA Netw Open. Aug 1 2023;6(8):e2326458. doi: 10.1001/jamanetworkopen.2023.26458 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Cholankeril G, Podboy A, Alshuwaykh OS, et al. Early Impact of COVID-19 on Solid Organ Transplantation in the United States. Transplantation. Nov 2020;104(11):2221–2224. doi: 10.1097/TP.0000000000003391 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Crooks PW, Thomas CO, Compton-Phillips A, et al. Clinical outcomes and healthcare use associated with optimal ESRD starts. Am J Manag Care. Oct 1 2018;24(10):e305–e311. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Cohort construction for analyses of the percentage of patients with optimal kidney failure treatment starts and related outcomes.
Figure S2. Unadjusted yearly percentages of optimal kidney failure treatment starts, overall and by component, by residence in states that did or did not expand Medicaid.
Figure S3. Unadjusted yearly percentages of kidney failure treatment starts with Medicaid, without insurance, and with prior nephrology care, by residence in states that did or did not expand Medicaid.
Figure S4. Cohort construction for analyses of the incidence of total and optimal kidney failure treatment starts in the general US population.
Figure S5. Adjusted yearly incidence of optimal and total kidney failure treatment starts by residence in states that did or did not expand Medicaid.
Figure S6. Adjusted yearly incidence of each optimal kidney failure treatment start component by residence in states that did or did not expand Medicaid.
Table S1. Fully-adjusted trends in percentage difference of optimal kidney failure treatment starts, overall and by component, by residence in states that did or did not expand Medicaid (sensitivity analysis treating 2014 and 2014–2015 as transition years).
Table S2. Fully-adjusted yearly percentage differences of kidney failure treatment starts with Medicaid, without insurance, and with prior nephrology care, by residence in states that did or did not expand Medicaid.
Data Availability Statement
The individual patient-level data underlying the results presented in this manuscript are freely available, subject to terms of a data use agreement with the United States Renal Data System and NIDDK, to qualified individuals or institutions.