Research ArticleORIGINAL RESEARCH
Open Access

Using Machine Learning to Identify Geographic and Socioeconomic Disparities in Dialysis Facility Outcomes Across the United States

Ziad M. Ashkar and Raju Gottumukkala
Ochsner Journal September 2025, 25 (3) 170-180; DOI: https://doi.org/10.31486/toj.25.0040

Abstract

Background Despite progress in dialysis care, the patient outcomes of mortality, hospitalization, and readmission rates remain unsatisfactory because of complex clinical, demographic, and socioeconomic interactions. For this study, we used unsupervised machine learning to identify clusters of dialysis facilities based on quality metrics and sociodemographic factors, with attention to racial and geographic disparities.

Methods We sourced facility-level data from data.cms.gov and sourced ZIP Code Tabulation Area–level sociodemographic data from the 2021 American Community Survey via the US Census Bureau application programming interface. Datasets were linked by ZIP code, standardized, and analyzed using principal component analysis and k-means clustering. We examined geographic patterns by US Census Bureau regions. Analyses were conducted in Python version 3.11.6 (Python Software Foundation) with the following libraries: pandas for data manipulation, scikit-learn for machine learning and principal component analysis, Matplotlib and Seaborn for data visualization, and GeoPandas for geographic mapping and spatial analysis.

Results Two facility clusters emerged: Cluster 0 (n=4,609) and Cluster 1 (n=2,857). Cluster 1 was characterized by poorer outcomes (higher mortality, hospitalization, readmission, anemia, catheter use, and hyperphosphatemia); lower rates of fistula use; and lower dialysis adequacy compared to Cluster 0. Cluster 1 facilities were more prevalent in regions with lower income, higher unemployment, and lower college education, and they served populations with greater proportions of Black and Hispanic residents. Geographically, Cluster 1 facilities were concentrated in the southern and western United States. Compared to Cluster 0, a larger share of Cluster 1 facilities were for-profit facilities (91.4% vs 88.5%).

Conclusion This study highlights a distinct cluster of underperforming dialysis clinics serving socioeconomically disadvantaged and racially diverse populations. Addressing these disparities requires multifaceted strategies including patient-level, institutional, and policy-level interventions.

Keywords:

INTRODUCTION

Dialysis outcomes—particularly mortality, hospitalization, and readmission rates—continue to pose major clinical and public health challenges. In the United States, annual mortality rates among patients with end-stage renal disease (ESRD) on hemodialysis exceed 20%.1 The high rates of hospitalization and readmission in dialysis patients increase health care costs and reflect persistent gaps in care quality.2,3 Traditional dialysis quality metrics such as Kt/V (a ratio that assesses the adequacy of dialysis treatment where K is dialyzer clearance [how effectively dialysis removes urea from the blood], t is the dialysis session time, and V is the volume of distribution [ie, the volume of body fluid where urea is distributed]); vascular access type; and treatment adherence are associated with adverse outcomes. For instance, the use of central venous catheters and poor adherence to treatment sessions correlate with elevated risks of hospitalization and death,4 while hyperphosphatemia has also been linked to increased mortality.5 However, clinical indicators alone do not fully capture patient risk.

Machine learning is a field of artificial intelligence that uses algorithms to identify patterns in data and to make predictions or classifications based on those patterns. In biomedical applications, machine learning methods are used to analyze complex, high-dimensional datasets.6

Supervised machine learning is a data-driven approach that leverages labeled clinical data to develop predictive or classification models and is the predominant machine learning paradigm used in current medical research. Supervised learning encompasses a range of techniques, from traditional regression models to more complex algorithms such as random forest and support vector machine.7 Unlike supervised learning, which requires labeled data (eg, known diagnoses), unsupervised learning operates on unlabeled data and is particularly valuable for hypothesis generation, exploratory data analysis, and precision medicine initiatives. These unsupervised learning algorithms autonomously identify hidden patterns, groupings, or structures within the data. Common methods include clustering (eg, k-means, hierarchical clustering), dimensionality reduction (eg, principal component analysis), and feature learning.8

Supervised machine learning algorithms, including random forest and support vector machine, have demonstrated superior accuracy in predicting readmission and mortality among dialysis patients compared to conventional models such as linear regression.9,10 Machine learning models have also been used to assess the impact of quality metrics such as dialysis adequacy and anemia management.11,12 Unsupervised machine learning has been used to identify subpopulations of dialysis patients who have unique risk profiles for complications such as intradialytic hypotension, anemia, or infection that may not be apparent using traditional clinical criteria.13

Sociodemographic and regional disparities remain pivotal to understanding variation in dialysis outcomes. Golestaneh et al demonstrated that patients receiving hemodialysis in communities with a higher percentage of Black residents had a significantly higher adjusted rate of hospitalization, independent of patient comorbidities and dialysis care factors.14 Wilkinson et al conducted a systematic review and found that ethnic minorities, including Black and Hispanic patients, face persistent disadvantages in dialysis adequacy, access to treatment modalities, and mortality, with these disparities persisting after adjustment for clinical variables.15

Factors such as socioeconomic status, health literacy, and geographic access to care contribute to these disparities,16 underscoring the necessity of integrating social determinants of health into predictive modeling and policy planning.17

Substantial evidence points to geographic disparities in dialysis outcomes across the United States. For instance, a seminal analysis of United States Renal Data System data by Wetmore et al (2017) found notable regional variations in ESRD mortality and hospitalization, with the Southeast exhibiting consistently higher mortality rates despite similar clinical practices elsewhere.18 These disparities persisted even after adjustment for demographics and comorbidities. Staffing differences also contribute to variations in outcomes. Plantinga et al showed that dialysis facilities with the highest patient to patient care technician ratios are associated with worse patient outcomes.19 In an analysis of 4,800 dialysis facilities, Yoder et al reported that the average registered nurse to patient ratio was highest in the Northeast and Midwest, intermediate in the West, and lowest in the South.20 Access to care and regional characteristics further modulate dialysis-related risks.21,22

For this study, we applied cluster analysis to explore dialysis outcomes across the 50 US states, with a particular focus on identifying sociodemographic and geographic disparities.

METHODS

Facility-level data were obtained from the Centers for Medicare & Medicaid Services (CMS) Dialysis Facility Compare (DFC) dataset, accessed via data.cms.gov, with a refresh date of January 2025. The analysis focused on the DFC_FACILITY file that includes facility-level mortality rates for the period January 1, 2020, to December 31, 2023. To ensure national representativeness, only facilities located in the 50 US states were included, resulting in a final sample of 7,488 dialysis facilities. Sociodemographic data at the ZIP Code Tabulation Area level were retrieved from the 2021 American Community Survey 5-year estimates via the US Census Bureau application programming interface. These data were linked to dialysis facility ZIP codes using a left join, achieving a 99.7% match rate (n=7,466 matched facilities).

Linked matched data (n=7,466 facilities) preprocessing was then conducted to ensure compatibility with machine learning workflows. Steps included (1) missing data handling–numerical columns with missing values were imputed using the mean of each respective feature; (2) feature scaling–numerical features were standardized to have zero mean and unit variance to ensure equal contribution across variables; and (3) dimensionality reduction–a total of 35 numerical features were included in a principal component analysis to reduce dimensionality while preserving variance. The number of components retained was determined by setting a 95% cumulative explained variance threshold. The resulting principal component analysis transformation retained 29 principal components, reflecting the dataset's high dimensionality and complexity. Principal component 1 (PC1) is the linear combination that captures the largest share of the variance in the dataset and is dominated by dialysis quality. Principal component 2 (PC2) captures the next share of variance and is dominated by demographics.

The principal component analysis–transformed data were subjected to k-means clustering to uncover natural groupings among facilities. The optimal number of clusters was determined using the silhouette score, with the number yielding the highest score selected as optimal. Final clustering was performed using k-means with this optimal cluster count.

Geographic distribution was analyzed across US states and census regions. Statistical comparisons between clusters were conducted using one-way analysis of variance for continuous variables and chi-square tests for categorical variables. All data processing, analysis, and visualization were conducted using Python version 3.11.6 (Python Software Foundation) with the following libraries: pandas for data manipulation, scikit-learn for machine learning and principal component analysis, Matplotlib and Seaborn for data visualization, and GeoPandas for geographic mapping and spatial analysis.

RESULTS

Cluster Analysis of Dialysis Facilities

Using principal component analysis and k-means clustering, 2 distinct groups of dialysis facilities were identified (Calinski-Harabasz score=505.83). Cluster 0 comprised 4,609 facilities (61.7%), and Cluster 1 comprised 2,857 facilities (38.3%). Figure 1 shows the cluster distribution. As discussed in the Methods section, PC1 is dominated by dialysis quality, and PC2 is dominated by demographics. Clear cluster separation is noted, supporting the validity of further analysis.

Figure 1.
Figure 1.

Cluster distribution of dialysis facilities using k-means clustering. Principal component 1 (PC1) is the linear combination that captures the largest share of the variance in the dataset and is dominated by dialysis quality. Principal component 2 (PC2) captures the next share of variance and is dominated by demographics.

Table 1 provides a summary of dialysis facility data, with variables categorized as categorical and continuous. The dataset includes facility characteristics, services offered, patient demographics, and key clinical performance metrics.

View this table:
Table 1.

Summary of Dialysis Facility Data, n=7,466

Table 2 shows dialysis facility characteristics by cluster for the continuous variables. Compared to Cluster 0 facilities, Cluster 1 facilities had higher mortality, hospitalization, and readmission rates and worse quality metrics: worse clearance (percent of patients with Kt/V >1.2), lower fistula use, and worse anemia (patients with hemoglobin <10 g/dL) and hyperphosphatemia (patients with serum phosphorus >7 mg/dL) management. Cluster 1 facilities also had a lower percentage of patients waitlisted for transplant. Moreover, higher percentages of Black and Hispanic patients were treated at Cluster 1 facilities, and overall, the patients at Cluster 1 facilities had lower household incomes, a lower percentage of college educations, and higher unemployment rates compared to patients at Cluster 0 facilities.

View this table:
Table 2.

Dialysis Facility Characteristics by Cluster, Continuous Variables

Table 3 shows dialysis facility characteristics by cluster for the categorical variables. Compared to Cluster 0, a higher percentage of Cluster 1 facilities—the cluster with worse mortality, hospitalization, and readmission outcomes—were for-profit and offered late shifts (late afternoon or early evening timing for dialysis). In addition, lower percentages of Cluster 1 facilities offered peritoneal dialysis and home hemodialysis training compared to Cluster 0 facilities.

View this table:
Table 3.

Dialysis Facility Characteristics by Cluster, Categorical Variables

To determine cluster distribution by region, the 50 US states were mapped by US Census Bureau region.23 Figure 2 shows the cluster distribution by region. The Midwest and Northeast regions have the highest proportion of Cluster 0 facilities (77.5% and 70.6%, respectively), while the South and West regions have the highest proportion of Cluster 1 facilities (46.0% and 44.6%, respectively).

Figure 2.
Figure 2.

Cluster 0 and Cluster 1 dialysis facility distribution by US Census Bureau region.

Figure 3 shows Cluster 1 facility distribution in the 48 contiguous states. States with high percentages of Cluster 1 facilities are principally located in the South and West. The 9 states with the highest percentages of Cluster 1 facilities are Texas (65%), California (62%), Georgia (62%), New Mexico (61%), Mississippi (57%), Nevada (54%), Louisiana (51%), Arizona (45%), and Florida (33%).

Figure 3.
Figure 3.

Percentages of Cluster 1 dialysis facilities by state. The 9 states with the highest percentages of Cluster 1 facilities are Texas (65%), California (62%), Georgia (62%), New Mexico (61%), Mississippi (57%), Nevada (54%), Louisiana (51%), Arizona (45%), and Florida (33%).

Table 4 shows economic indicators by US Census Bureau region. The lowest median household income and lowest rate of college education are in the South.

View this table:
Table 4.

Economic Indicators by US Census Bureau Region Where Dialysis Facilities Are Located

Figure 4 shows the racial/ethnic distribution by US Census Bureau region of the population where the dialysis clinics are located. The highest percentage of Blacks (24.5%) is in the South, and the highest percentages of Asians (10.8%) and Hispanics/Latinos (33.2%) are in the West. The highest percentages of Whites are in the Midwest (69.0%) and Northeast (60.2%).

Figure 4.
Figure 4.

Racial/ethnic composition comparison across US Census Bureau regions.

Table 5 shows dialysis facility quality indicators (outcomes) by US Census Bureau region. The South has the highest mortality rate, lowest fistula rate, highest anemia rate, highest hyperphosphatemia rate, highest percentage of for-profit facilities, and the lowest percentage of facilities offering home hemodialysis training. The West has the second highest mortality rate, the second highest hyperphosphatemia rate, and second lowest percentage of facilities offering home hemodialysis training. Noticeably, the West has the highest percentage of facilities offering late shifts and highest number of dialysis stations.

View this table:
Table 5.

Dialysis Facility Quality Indicators by US Census Bureau Region

DISCUSSION

For this study, we used unsupervised machine learning to analyze data from CMS dialysis facilities and the US Census Bureau. Facilities in Cluster 1 exhibited worse clinical outcomes, including higher mortality, hospitalization, and readmission rates than facilities in Cluster 0. These outcomes were associated with lower quality metrics, such as suboptimal metrics in dialysis clearance (Kt/V ≥1.2), mineral bone disease (serum phosphorus >7.0 mg/dL), anemia (hemoglobin <10 g/dL), and vascular access (fistula rate and long-term catheter use).

These findings are consistent with prior research. For instance, anemia, particularly when severe, is a known predictor of mortality and hospitalization during maintenance dialysis.24,25 Lacson et al showed that dialysis facilities with better performance in vascular access use, anemia management, and dialysis clearance demonstrated lower mortality and hospitalization rates.26 Suboptimal control of mineral metabolism—such as abnormally high phosphorus, calcium, or parathyroid hormone levels—has also been linked to increased cardiovascular mortality and hospitalizations.27 In our study, we found higher percentages of hyperphosphatemia (phosphorus >7 mg/dL) and hypercalcemia (calcium >10.2 mg/dL) in Cluster 1 compared to Cluster 0; however, noticeably, Cluster 1 had a higher percentage of hypophosphatemia (phosphorus <3.5 mg/dL) than Cluster 0. Hypophosphatemia has been linked to increased mortality in dialysis patients, and has also been associated with hypoalbuminemia, reflecting poor nutritional and functional status.28,29

Cluster 1 facilities had a lower percentage of patients aged 65 years or older than Cluster 0 facilities. Villar et al showed that older patients with ESRD experienced less excess mortality than younger patients with ESRD.30 Kucirka et al compared mortality among races and showed that Black patients on dialysis had a lower risk of death compared to White patients, but this survival advantage only applied to older adults (>50 years).31 Of note, the Cluster 1 facilities in our study had a higher percentage of Black patients than the Cluster 0 facilities.

Compared to Cluster 0 facilities, Cluster 1 facilities were also notable for a lower percentage of patients on the transplant waitlist, a factor known to be associated with increased mortality in dialysis.32 Compared to Cluster 0 facilities, Cluster 1 facilities also had a higher proportion of for-profit clinics, a finding that has been associated with adverse outcomes, including higher rates of mortality and hospitalization.33,34 Contributing factors include staffing, resource allocation, financial pressures, and the likelihood of referral for kidney transplant.35

In our study, facilities in Cluster 1 had a larger average number of dialysis stations and a greater likelihood of offering late shifts. Studies have shown an increased mortality risk in the smallest facilities (≤15 stations) but no demonstrated difference in mortality or quality outcomes in facilities with 16 to 20 stations vs facilities with >20 stations.36,37

Peritoneal dialysis and home hemodialysis have been associated with improved patient outcomes, including reduced mortality and hospitalization.38-40 In their systematic review, Walker et al found that patients on home hemodialysis reported satisfaction with staff expertise and educational support.41 In our study, a lower percentage of Cluster 1 facilities offered peritoneal dialysis and home hemodialysis training compared to Cluster 0 facilities.

Mapping by US Census Bureau regions revealed that Cluster 1 facilities were principally concentrated in the South and West. State-level analysis further confirmed that states in the South and West had the highest proportions of Cluster 1 facilities. Some of our findings align with previous data that report higher dialysis mortality and hospitalization rates in the South compared to other US regions,42 although mortality in the West has been reported to be lower than in the South and lower to comparable compared to mortality in the Midwest and Northeast.43 To further understand these potential conflicting differences—given that we showed higher mortality in the West—we looked at the sociodemographic analysis of the US Census Bureau regions. These data showed that the South had the lowest median household income and college education rate among the 4 regions. The South also had the highest proportion of Blacks in the covered ZIP codes of the dialysis service areas, while the West had the highest proportion of Hispanics in the covered ZIP codes of the dialysis service areas. As noted previously, prior studies have demonstrated strong associations between worse health outcomes and socioeconomic disadvantage and racial minority status.14,15 When we compared regional quality metrics and other dialysis facility characteristics by US Census Bureau region, we found that the South had the highest mortality rate, lowest fistula rate, highest anemia rate, highest hyperphosphatemia rate, highest percent of for-profit facilities, and lowest percent of facilities offering home hemodialysis training. In comparison, the West had the second highest mortality rate, second highest hyperphosphatemia rate, and the second lowest percentage of clinics offering home hemodialysis training. Together, these factors—clinical performance, facility characteristics, and sociodemographic context—provide a reasonable explanation for the geographic concentration of Cluster 1 facilities in the South and West.

Based on our study findings, there is a great need for multidisciplinary interventions focused on care access, resource allocation, and dialysis quality metrics to yield improvements in dialysis patient outcomes. Further research should incorporate patient-level comorbidity data, staffing models, and regional policy analysis to further refine our understanding of mortality and quality disparities in ESRD care. Ultimately, addressing these inequities will require targeted clinical, operational, and policy-level reforms to ensure equitable access to high-quality dialysis care across all communities.

This study has several limitations. First, as an unsupervised machine learning study, clustering techniques rely on input variables, and some outcome predictors may not have been captured. Second, our analysis is based on facility-level data that may not fully reflect individual patient characteristics or comorbidities, such as nutritional status and diabetes. Third, confounding factors such as local health care policies, insurance coverage, and referral practices were not directly analyzed. Strengths of the study are its national scope and integration of facility quality metrics with community-level sociodemographic data. The unsupervised machine learning approach allowed for data-driven discovery of latent facility groupings not constrained by a priori assumptions.

CONCLUSION

This study used unsupervised machine learning to identify clusters of dialysis facilities with distinct quality profiles, uncovering significant disparities in mortality, hospitalization, and readmission rates. Facilities in Cluster 1—representing nearly 40% of the national sample—were disproportionately located in socioeconomically disadvantaged areas—predominantly in the South and West regions of the United States—and served racially diverse populations. These facilities exhibited poorer performance across key quality indicators, including vascular access, dialysis adequacy, anemia management, hyperphosphatemia, and transplant waitlisting compared to the Cluster 0 facilities. The findings underscore a pressing need for multifaceted interventions including allocating resources to high-risk regions, implementing programs focused on improving quality processes (such as vascular access, mineral bone disease control, and anemia management), expanding access to home dialysis training, and enhancing transplant waitlist processes in underserved areas. Addressing these factors could potentially improve patient outcomes. Our study highlights the utility of machine learning approaches in uncovering systemic patterns in health care delivery. By integrating facility-level quality metrics with sociodemographic context, this study offers a novel framework for identifying geographic and structural disparities in dialysis care.

This article meets the Accreditation Council for Graduate Medical Education and the American Board of Medical Specialties Maintenance of Certification competencies for Patient Care and Medical Knowledge.

ACKNOWLEDGMENTS

The authors have no financial or proprietary interest in the subject matter of this article.

©2025 by the author(s); licensee Ochsner Journal, Ochsner Clinic Foundation, New Orleans, LA. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (creativecommons.org/licenses/by/4.0/legalcode) that permits unrestricted use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

REFERENCES

  1. 1.↵
    National Institute of Diabetes and Digestive and Kidney Diseases, United States Renal Data System. 2024 Annual Data Report. U.S. Department of Health and Human Services, National Institutes of Health. Accessed July 29, 2025. usrds-adr.niddk.nih.gov/2024
  2. 2.↵
    1. Chan L,
    2. Chauhan K,
    3. Poojary P,
    4. et al.
    National estimates of 30-day unplanned readmissions of patients on maintenance hemodialysis. Clin J Am Soc Nephrol. 2017;12(10):1652-1662. doi: 10.2215/CJN.02600317
    OpenUrlAbstract/FREE Full Text
  3. 3.↵
    1. Mathew AT,
    2. Strippoli GF,
    3. Ruospo M,
    4. Fishbane S
    . Reducing hospital readmissions in patients with end-stage kidney disease. Kidney Int. 2015;88(6):1250-1260. doi: 10.1038/ki.2015.307
    OpenUrlCrossRefPubMed
  4. 4.↵
    1. Rocco MV,
    2. Frankenfield DL,
    3. Hopson SD,
    4. McClellan WM
    . Relationship between clinical performance measures and outcomes among patients receiving long-term hemodialysis. Ann Intern Med. 2006;145(7):512-519. doi: 10.7326/0003-4819-145-7-200610030-00009
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Block GA,
    2. Hulbert-Shearon TE,
    3. Levin NW,
    4. Port FK
    . Association of serum phosphorus and calcium x phosphate product with mortality risk in chronic hemodialysis patients: a national study. Am J Kidney Dis. 1998;31(4):607-617. doi: 10.1053/ajkd.1998.v31.pm9531176
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Binson VA,
    2. Thomas S,
    3. Subramoniam M,
    4. Arun J,
    5. Naveen S,
    6. Madhu S
    . A review of machine learning algorithms for biomedical applications. Ann Biomed Eng. 2024;52(5):1159-1183. doi: 10.1007/s10439-024-03459-3
    OpenUrlCrossRefPubMed
  7. 7.↵
    1. Wiemken TL,
    2. Kelley RR
    . Machine learning in epidemiology and health outcomes research. Annu Rev Public Health. 2020;41:21-36. doi: 10.1146/annurev-publhealth-040119-094437
    OpenUrlCrossRefPubMed
  8. 8.↵
    1. Eckhardt CM,
    2. Madjarova SJ,
    3. Williams RJ,
    4. et al.
    Unsupervised machine learning methods and emerging applications in healthcare. Knee Surg Sports Traumatol Arthrosc. 2023;31(2):376-381. doi: 10.1007/s00167-022-07233-7
    OpenUrlCrossRefPubMed
  9. 9.↵
    1. Garcia-Montemayor V,
    2. Martin-Malo A,
    3. Barbieri C,
    4. et al.
    Predicting mortality in hemodialysis patients using machine learning analysis. Clin Kidney J. 2020;14(5):1388-1395. doi: 10.1093/ckj/sfaa126
    OpenUrlCrossRefPubMed
  10. 10.↵
    1. Tsai CH,
    2. Shih DH,
    3. Tu JH,
    4. Wu TW,
    5. Tsai MG,
    6. Shih MH
    . Analyzing monthly blood test data to forecast 30-day hospital readmissions among maintenance hemodialysis patients. J Clin Med. 2024;13(8):2283. doi: 10.3390/jcm13082283
    OpenUrlCrossRefPubMed
  11. 11.↵
    1. Kim HW,
    2. Heo SJ,
    3. Kim JY,
    4. Kim A,
    5. Nam CM,
    6. Kim BS
    . Dialysis adequacy predictions using a machine learning method. Sci Rep. 2021;11(1):15417. doi: 10.1038/s41598-021-94964-1
    OpenUrlCrossRefPubMed
  12. 12.↵
    1. Kang C,
    2. Han J,
    3. Son S,
    4. et al.
    Optimizing anemia management using artificial intelligence for patients undergoing hemodialysis. Sci Rep. 2024;14(1):26739. doi: 10.1038/s41598-024-75995-w
    OpenUrlCrossRefPubMed
  13. 13.↵
    1. Burlacu A,
    2. Iftene A,
    3. Jugrin D,
    4. et al.
    Using artificial intelligence resources in dialysis and kidney transplant patients: a literature review. Biomed Res Int. 2020;2020:9867872. doi: 10.1155/2020/9867872
    OpenUrlCrossRefPubMed
  14. 14.↵
    1. Golestaneh L,
    2. Cavanaugh KL,
    3. Lo Y,
    4. et al.
    Community racial composition and hospitalization among patients receiving in-center hemodialysis. Am J Kidney Dis. 2020;76(6):754-764. doi: 10.1053/j.ajkd.2020.05.019
    OpenUrlCrossRefPubMed
  15. 15.↵
    1. Wilkinson E,
    2. Brettle A,
    3. Waqar M,
    4. Randhawa G
    . Inequalities and outcomes: end stage kidney disease in ethnic minorities. BMC Nephrol. 2019;20(1):234. doi: 10.1186/s12882-019-1410-2
    OpenUrlCrossRefPubMed
  16. 16.↵
    1. Taylor DM,
    2. Fraser S,
    3. Dudley C,
    4. et al.
    Health literacy and patient outcomes in chronic kidney disease: a systematic review. Nephrol Dial Transplant. 2018;33(9):1545-1558. doi: 10.1093/ndt/gfx293
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Purtell L,
    2. Bennett P,
    3. Bonner A
    . Multimodal approaches for inequality in kidney care: turning social determinants of health into opportunities. Curr Opin Nephrol Hypertens. 2024;33(1):34-42. doi: 10.1097/MNH.0000000000000936
    OpenUrlCrossRefPubMed
  18. 18.↵
    1. Wetmore JB,
    2. Liu J,
    3. Li S,
    4. et al.
    The Healthy People 2020 Objectives for Kidney Disease: how far have we come, and where do we need to go? Clin J Am Soc Nephrol. 2017;12(1):200-209. doi: 10.2215/CJN.04210416
    OpenUrlAbstract/FREE Full Text
  19. 19.↵
    1. Plantinga LC,
    2. Bender AA,
    3. Urbanski M,
    4. et al.
    Patient care technician staffing and outcomes among US patients receiving in-center hemodialysis. JAMA Netw Open. 2024;7(3):e241722. doi: 10.1001/jamanetworkopen.2024.1722 [published correction appears in JAMA Netw Open. 2024;7(5):e2419682. doi: 10.1001/jamanetworkopen.2024.19682]
    OpenUrlCrossRef
  20. 20.↵
    1. Yoder LA,
    2. Xin W,
    3. Norris KC,
    4. Yan G
    . Patient care staffing levels and facility characteristics in U.S. hemodialysis facilities. Am J Kidney Dis. 2013;62(6):1130-1140. doi: 10.1053/j.ajkd.2013.05.007
    OpenUrlCrossRefPubMed
  21. 21.↵
    1. Mehrotra R,
    2. Story K,
    3. Guest S,
    4. Fedunyszyn M
    . Neighborhood location, rurality, geography, and outcomes of peritoneal dialysis patients in the United States. Perit Dial Int. 2012;32(3):322-331. doi: 10.3747/pdi.2011.00084
    OpenUrlAbstract/FREE Full Text
  22. 22.↵
    1. Bowe B,
    2. Xie Y,
    3. Xian H,
    4. Lian M,
    5. Al-Aly Z
    . Geographic variation and US county characteristics associated with rapid kidney function decline. Kidney Int Rep. 2016;2(1):5-17. doi: 10.1016/j.ekir.2016.08.016
    OpenUrlCrossRefPubMed
  23. 23.↵
    Census regions and divisions of the United States. U.S. Census Bureau. Accessed June 22, 2025. www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf
  24. 24.↵
    1. Portolés J,
    2. López-Gómez JM,
    3. Aljama P
    . A prospective multicentre study of the role of anaemia as a risk factor in haemodialysis patients: the MAR Study. Nephrol Dial Transplant. 2007;22(2):500-507. doi: 10.1093/ndt/gfl558
    OpenUrlCrossRefPubMed
  25. 25.↵
    1. Robinson BM,
    2. Joffe MM,
    3. Berns JS,
    4. Pisoni RL,
    5. Port FK,
    6. Feldman HI
    . Anemia and mortality in hemodialysis patients: accounting for morbidity and treatment variables updated over time. Kidney Int. 2005;68(5):2323-2330. doi: 10.1111/j.1523-1755.2005.00693.x
    OpenUrlCrossRefPubMed
  26. 26.↵
    1. Lacson E Jr.,
    2. Wang W,
    3. Lazarus JM,
    4. Hakim RM
    . Hemodialysis facility-based quality-of-care indicators and facility-specific patient outcomes. Am J Kidney Dis. 2009;54(3):490-497. doi: 10.1053/j.ajkd.2009.01.260
    OpenUrlCrossRefPubMed
  27. 27.↵
    1. Block GA,
    2. Kilpatrick RD,
    3. Lowe KA,
    4. Wang W,
    5. Danese MD
    . CKD-mineral and bone disorder and risk of death and cardiovascular hospitalization in patients on hemodialysis. Clin J Am Soc Nephrol. 2013;8(12):2132-2140. doi: 10.2215/CJN.04260413
    OpenUrlAbstract/FREE Full Text
  28. 28.↵
    1. Lertdumrongluk P,
    2. Rhee CM,
    3. Park J,
    4. et al.
    Association of serum phosphorus concentration with mortality in elderly and nonelderly hemodialysis patients. J Ren Nutr. 2013;23(6):411-421. doi: 10.1053/j.jrn.2013.01.018
    OpenUrlCrossRefPubMed
  29. 29.↵
    1. Lee JE,
    2. Lim JH,
    3. Jang HM,
    4. et al.
    Low serum phosphate as an independent predictor of increased infection-related mortality in dialysis patients: a prospective multicenter cohort study. PLoS One. 2017;12(10):e0185853. doi: 10.1371/journal.pone.0185853
    OpenUrlCrossRefPubMed
  30. 30.↵
    1. Villar E,
    2. Remontet L,
    3. Labeeuw M,
    4. Ecochard R
    . Effect of age, gender, and diabetes on excess death in end-stage renal failure. J Am Soc Nephrol. 2007;18(7):2125-2134. doi: 10.1681/ASN.2006091048
    OpenUrlAbstract/FREE Full Text
  31. 31.↵
    1. Kucirka LM,
    2. Grams ME,
    3. Lessler J,
    4. et al.
    Association of race and age with survival among patients undergoing dialysis. JAMA. 2011;306(6):620-626. doi: 10.1001/jama.2011.1127
    OpenUrlCrossRefPubMed
  32. 32.↵
    1. Wolfe RA,
    2. Ashby VB,
    3. Milford EL,
    4. et al.
    Comparison of mortality in all patients on dialysis, patients on dialysis awaiting transplantation, and recipients of a first cadaveric transplant. N Engl J Med. 1999;341(23):1725-1730. doi: 10.1056/NEJM199912023412303
    OpenUrlCrossRefPubMed
  33. 33.↵
    1. Dickman S,
    2. Mirza R,
    3. Kandi M,
    4. et al.
    Mortality at for-profit versus not-for-profit hemodialysis centers: a systematic review and meta-analysis. Int J Health Serv. 2021;51(3):371-378. doi: 10.1177/0020731420980682
    OpenUrlCrossRefPubMed
  34. 34.↵
    1. Dalrymple LS,
    2. Johansen KL,
    3. Romano PS,
    4. et al.
    Comparison of hospitalization rates among for-profit and nonprofit dialysis facilities. Clin J Am Soc Nephrol. 2014;9(1):73-81. doi: 10.2215/CJN.04200413
    OpenUrlAbstract/FREE Full Text
  35. 35.↵
    1. Garg PP,
    2. Frick KD,
    3. Diener-West M,
    4. Powe NR
    . Effect of the ownership of dialysis facilities on patients' survival and referral for transplantation. N Engl J Med. 1999;341(22):1653-1660. doi: 10.1056/NEJM199911253412205
    OpenUrlCrossRefPubMed
  36. 36.↵
    1. Yan G,
    2. Norris KC,
    3. Xin W,
    4. et al.
    Facility size, race and ethnicity, and mortality for in-center hemodialysis. J Am Soc Nephrol. 2013;24(12):2062-2070. doi: 10.1681/ASN.2013010033
    OpenUrlAbstract/FREE Full Text
  37. 37.↵
    1. Ajmal F,
    2. Probst J,
    3. Brooks J,
    4. Hardin J,
    5. Qureshi Z
    . Association between freestanding dialysis facility size and Medicare Quality Incentive Program performance scores. Am J Nephrol. 2019;49(1):64-73. doi: 10.1159/000495262
    OpenUrlCrossRefPubMed
  38. 38.↵
    1. Marshall MR,
    2. Polkinghorne KR,
    3. Boudville N,
    4. McDonald SP
    . Home versus facility dialysis and mortality in Australia and New Zealand. Am J Kidney Dis. 2021;78(6):826-836.e1. doi: 10.1053/j.ajkd.2021.03.018
    OpenUrlCrossRefPubMed
  39. 39.
    1. Yeung EK,
    2. Polkinghorne KR,
    3. Kerr PG
    . Home and facility haemodialysis patients: a comparison of outcomes in a matched cohort. Nephrol Dial Transplant. 2021;36(6):1070-1077. doi: 10.1093/ndt/gfaa358
    OpenUrlCrossRefPubMed
  40. 40.↵
    1. Sarnak MJ,
    2. Auguste BL,
    3. Brown E,
    4. et al.
    Cardiovascular effects of home dialysis therapies: a scientific statement from the American Heart Association. Circulation. 2022;146(11):e146-e164. doi: 10.1161/CIR.0000000000001088
    OpenUrlCrossRefPubMed
  41. 41.↵
    1. Walker RC,
    2. Hanson CS,
    3. Palmer SC,
    4. et al.
    Patient and caregiver perspectives on home hemodialysis: a systematic review. Am J Kidney Dis. 2015;65(3):451-463. doi: 10.1053/j.ajkd.2014.10.020
    OpenUrlCrossRefPubMed
  42. 42.↵
    1. Weinhandl ED,
    2. Ray D,
    3. Kubisiak KM,
    4. Collins AJ
    . Contemporary trends in clinical outcomes among dialysis patients with Medicare coverage. Am J Nephrol. 2019;50(1):63-71. doi: 10.1159/000500943
    OpenUrlCrossRefPubMed
  43. 43.↵
    1. Zarkowsky DS,
    2. Hicks CW,
    3. Arhuidese I,
    4. et al.
    Quality improvement targets for regional variation in surgical end-stage renal disease care. JAMA Surg. 2015;150(8):764-770. doi: 10.1001/jamasurg.2015.1126
    OpenUrlCrossRefPubMed
Back to top

In this issue

Print
Download PDF
Email Article
Citation Tools

Jump to section

Keywords