Dialysis Costs for a Health System Participating in Value-Based Care

ABSTRACT

Objectives: Unplanned “crash” dialysis starts are associated with worse outcomes and higher costs, a challenging problem for health systems participating in value-based care (VBC). We examined expenditures and utilization associated with these events in a large health system.

Study Design: Retrospective, single-center study at Cleveland Clinic, a large, integrated health system participating in VBC contracts, including a Medicare accountable care organization.

Methods: We analyzed beneficiaries who transitioned to dialysis between 2017 and 2020. Crash starts involved initiating inpatient hemodialysis (HD) with a central venous catheter (CVC). Optimal starts were initiated with either home dialysis or outpatient HD without a CVC. Suboptimal starts were initiated with outpatient HD with a CVC or inpatient HD without a CVC.

Results: A total of 495 patients initiated chronic dialysis: 260 crash starts, 130 optimal starts, and 105 suboptimal starts. Median predialysis 12-month cost was $67,059 for crash starts, $17,891 for optimal starts, and $7633 for suboptimal starts (P < .001). Median postdialysis 12-month cost was $71,992 for crash starts, $55,427 for optimal starts, and $72,032 for suboptimal starts (P = .001). Predialysis inpatient admission per 1000 beneficiaries was 1236 per 1000 for crash starts vs 273 per 1000 for optimal starts and 170 per 1000 for suboptimal starts (P < .001). Postdialysis inpatient admission for crash starts was 853 per 1000 vs 291 per 1000 for optimal starts and 184 per 1000 for suboptimal starts (P < .001).

Conclusions: In a major health system, crash starts demonstrated the highest cost and hospital utilization, a pattern that persisted after dialysis initiation. Developing strategies to promote optimal starts will improve VBC contract performance.

Am J Manag Care. 2023;29(8):e235-e241. https://doi.org/10.37765/ajmc.2023.89410

Takeaway Points

Understanding the different categories and consequences of dialysis initiation is critical for health systems in value-based care (VBC). We observed significantly higher costs and utilization for patients with unplanned “crash” vs planned “optimal” dialysis initiation. Health systems need strategies to promote optimal dialysis starts to improve VBC performance and reduce potentially avoidable outcomes.

• Crash starts had higher expenditures and hospital utilization before and after chronic dialysis initiation compared with optimal starts.
• Nephrology and vascular surgery utilization was significantly different among categories and highest for optimal starts.
• Suboptimal dialysis starts, similar to crash starts, demonstrated high postdialysis costs.

Health systems across the United States are transitioning from fee for service to value-based care (VBC) and responsibility for total health care costs.^1,2 Accountable care organizations (ACOs) are now present in many health systems. In 2020, more than 11 million beneficiaries were enrolled in Medicare ACOs, with $2.3 billion in shared savings generated.^1,3 Other health system VBC arrangements include Medicare Advantage (MA) and commercial insurance plans.^1,2 Care for beneficiaries on dialysis with end-stage kidney disease (ESKD) and chronic kidney disease (CKD) imposes a large financial burden on payers.⁴ Optimizing utilization related to these conditions is essential for VBC and reducing avoidable expenditures.

The percentage of patients on dialysis attributed to Medicare ACOs has increased dramatically, rising from 6% to 23% in from 2012 to 2016 alone.⁵ These beneficiaries incur $51 billion in Medicare annual spending, with per-member per-month (PMPM) costs 6 to 7 times higher than average.⁴ In a VBC setting, addressing high-cost conditions is important for health system financial strategy.^5,6 The resource burden imposed by dialysis has spurred national attention on ESKD and kidney-specific payment models.^7-11 Following success of the dialysis-only Comprehensive End-Stage Renal Disease Care Model, Medicare launched the Kidney Care Choices (KCC) Innovation Model in 2022, expanding clinical and financial program scope to late-stage CKD in addition to dialysis.^7-9

The transition from CKD to dialysis is a critical target for quality and cost improvement.^12-14 Unplanned initiation of dialysis, or “crash start,” involves initiating hemodialysis (HD) with a central venous catheter (CVC) as an inpatient (IP). Crash starts are associated with higher risks of death, bloodstream infections, and rehospitalization.^4,10 Conversely, planned dialysis initiation, or “optimal start,” is associated with superior clinical outcomes and costs.^8,15,16 Optimal start is a quality measure endorsed by the National Quality Forum (NQF) and is a central focus in Medicare’s new KCC Innovation Model.^8,15-17 Optimal start is a composite outcome of either home dialysis, HD with a permanent vascular access (an arteriovenous fistula or arteriovenous graft [AVF/G]), or preemptive kidney transplantation.¹⁷ Kaiser Permanente, a major health system and payer, demonstrated the VBC benefits of implementing a large-scale optimal start program in its physician networks.^15,17,18

Cleveland Clinic Foundation (CCF) is a large multihospital health system and clinically integrated community practice network located in Northeast Ohio. CCF participates in a Medicare ACO, MA, and numerous commercial plans with attribution for more than 445,000 beneficiaries. With transition to downside ACO financial risk, reducing the high cost of dialysis was identified as a quality improvement (QI) priority. We performed this study to examine the financial and clinical features of dialysis initiation and understand their impact on VBC contract performance.

METHODS

A retrospective analysis of CCF VBC contracts between 2017 and 2020 was performed, a period chosen by virtue of having complete claims data. Our query included Medicare ACO, MA, and commercial VBC contracts. Patients with incomplete claims history were excluded, along with Medicaid and Veterans Affairs beneficiaries, due to small numbers. Beneficiaries who transitioned to dialysis were identified using their first outpatient dialysis claim as the reference point for analysis. An algorithm (eAppendix Figure [eAppendix available at ajmc.com]) using the reverse sequence of dialysis Common Procedural Terminology (CPT) codes (eAppendix Table 1) was used to categorize beneficiaries into different dialysis start categories. A secondary test using the forward sequence of claims was used to validate assignments, with adjudication of any discrepancies by the authors.

In addition to the previously defined optimal and crash starts, we created a novel category, “suboptimal start,” to describe patients initiating either outpatient HD using a CVC or IP HD using an AVF/G. All beneficiary IP and ambulatory claims and utilization were analyzed 12 months before (predialysis) and after (post dialysis) their first outpatient dialysis claim. This interval was selected to restrict analysis to utilization reasonably proximate to dialysis initiation. Patients who died or transferred out of CCF following dialysis initiation were included up to their last day of payer attribution. Patients requiring dialysis for acute kidney injury without an outpatient dialysis claim and/or who died after starting IP dialysis were excluded. Very few patients initiated PD in the hospital (eg, urgent start PD) and no patients initiated home hemodialysis, so they were excluded (data not shown). Although optimal start includes preemptive kidney transplantation, this population was out of scope for our QI activities and not examined.

SNOMED CT and our Epic Healthy Planet electronic registry were used to identify individual patient characteristics and variables associated with predialysis care. Information included International Classification of Diseases, Tenth Revision codes from the Epic problem list, professional and procedural CPT codes, site of service (IP, emergency department [ED], or ambulatory), specialty department, laboratory results, blood pressure (BP) values, medication lists, and date of death, if applicable. The last ambulatory data prior to the IP admission associated with dialysis initiation or the first outpatient dialysis claim were used, whichever came first. Predialysis clinical parameters were analyzed, including comorbidities and medications. We created a surrogate measure for optimal predialysis CKD care, defined as at least 2 ambulatory nephrology visits, BP less than 140/90 mm Hg, and treatment with an angiotensin-converting enzyme inhibitor (ACEI) or angiotensin receptor blocker (ARB). IP admission, IP readmission, and ED utilization rates were also calculated.

Statistical tests for differences between group characteristics and observed utilization were performed using analysis of variance (ANOVA) and t tests or Kruskal-Wallis 1-way ANOVA and Mann-Whitney U tests depending on data distributions, and χ² test or Fisher exact test as appropriate. Mean and SD or median and IQR, counts, and percentages were reported. For each rate of utilization for which only a single population value could be calculated, a bootstrap resampling method with 1000 replicates was used for estimating its CI and statistical testing. Univariate and multivariable logistic regression models were created to examine the relationship between IP and ED utilization and different subgroups, patient demographics, and clinical parameters. Odds ratios (ORs) and 95% CIs were calculated for all model variables.

This activity was granted a minimal-risk study exemption (IRB 21-593) by the Cleveland Clinic Foundation Institutional Review Board on May 19, 2021, as a QI activity. All data analysis and records review followed CCF and Medicare compliance and privacy policies.

RESULTS

Table 1 shows patient characteristics and ambulatory utilization in the 12 months prior to the first outpatient dialysis treatment. A total of 495 patients began chronic dialysis. Most were crash starts (52.5%), followed by optimal (26.3%) and suboptimal (21.2%) starts. Most optimal starts were outpatient with an AVF/G (72.3%) vs PD (27.7%) (data not shown). Payers were Medicare ACO (63%), followed by MA (30.1%) and commercial VBC (6.9%) (P = .008). Aside from patients with crash and optimal starts being slightly older (P = .001), there were no demographic differences. Crash starts demonstrated the highest primary care provider (PCP) utilization vs optimal or suboptimal starts (P < .001). Optimal starts had the highest nephrology and vascular surgery utilization (63.8% and 46.9% with any visit, respectively) and mean visits (3.24 and 0.81, respectively) vs crash or suboptimal starts (P < .001). Suboptimal starts displayed the lowest overall ambulatory PCP or specialist utilization. Optimal CKD care (as defined earlier) was low overall (9.1%) but highest (12.7%) for crash starts (P = .002).

Mean last estimated glomerular filtration rate (eGFR) for crash starts (13 mL/min/1.73 m²) was significantly higher than that for optimal (10.3 mL/min/1.73 m²) or suboptimal (10 mL/min/1.73 m²) starts (P < .001). Measurement of eGFR within 3 months and 6 months of dialysis initiation was similar between the crash and optimal start groups, with suboptimal starts having the lowest frequency of measurement (P < .001). Mean systolic BP was lowest in crash starts (P = .002). Congestive heart failure (CHF) was more prevalent in crash (61.9%) and suboptimal (51.4%) starts than optimal starts (44.6%) (P = .004). There was low use of heart- and kidney-protective ACEIs or ARBs (39%) in all groups; use was highest in the crash start (45.8%) compared with the optimal start (30.8%) or suboptimal start (32.4%) groups (P = .005). Other BP medications are listed in Table 1, showing only higher α-blocker use (30%) in optimal starts (P < .001). Median postdialysis TCOC for crash starts ($71,992) remained higher than for optimal starts ($55,427) although the margin narrowed (P = .001). The suboptimal start group had the lowest predialysis TCOC but highest postdialysis costs overall ($72,032). Postdialysis TCOC was truncated for 54 (10.9%) patients who died in the postdialysis period.

PMPM cost values were calculated and reflect costs up to death (Figure [B]). PMPM cost patterns mirrored the TCOC analysis. Crash starts had the highest predialysis median PMPM costs ($5588) compared with optimal ($1491) and suboptimal ($636) starts (P = .001). Postdialysis median PMPM costs remained highest for crash ($6974) vs optimal ($4627) starts, although suboptimal start PMPM costs rose substantially ($6109) (P < .001). High-cost outliers created a leftward-skewed distribution (Figure [B]). There were no significant differences in TCOC or PMPM cost between optimal start AVF/G or PD subgroups (data not shown).

Table 2 lists IP admission, IP readmission, and ED utilization rates per 1000 beneficiaries with 95% CIs. Crash starts had the highest predialysis (1236 per 1000) and postdialysis (853 per 1000) IP admission rates (P < .001). Crash start IP readmission rates predialysis (164 per 1000) and post dialysis (131 per 1000) were also the highest among the start groups (P < .001). ED utilization followed similar patterns, as crash start predialysis (392 per 1000) and postdialysis (402 per 1000) rates were the highest (P < .001). Although IP and ED utilization for optimal starts was much lower than that for crash starts, suboptimal starts displayed the lowest rates across all predialysis and postdialysis categories (P < .001).

Univariate and multivariable logistic regression models were constructed to examine relationships between hospital utilization rates and different clinical and demographic variables in the predialysis and postdialysis periods (eAppendix Table 2 [A-F]). Predialysis, the multivariable model demonstrated ORs for IP admission to be significantly associated with optimal start (OR, 0.20; 95% CI, 0.12-0.33), suboptimal start (OR, 0.16; 95% CI, 0.09-0.27), male gender (OR, 0.59; 95% CI, 0.38-0.90), hypertension (OR, 4.10; 95% CI, 1.73-9.75), and CHF (OR, 2.87; 95% CI, 1.87-4.39). The predialysis IP readmission multivariable model revealed suboptimal start (OR, 0.28; 95% CI, 0.09-0.87), higher hemoglobin A_1c (OR, 1.34; 95% CI, 1.09-1.64), and CHF (OR, 4.23; 95% CI, 1.91-9.35) as significant. For predialysis ED utilization, suboptimal start (OR, 0.43; 95% CI, 0.25-0.75), higher diastolic BP (OR, 1.02; 95% CI, 1.00-1.03), and peripheral vascular disease (OR, 1.65; 95% CI, 1.05-2.60) were identified.

Multivariable logistic regression analysis of postdialysis IP admission revealed associations with optimal start (OR, 0.56; 95% CI, 0.35-0.89), suboptimal start (OR, 0.48; 95% CI, 0.28-0.81), higher eGFR (OR, 0.97; 95% CI, 0.97-1.00), diabetes (OR, 1.67; 95% CI, 1.10-2.53), and CHF (OR, 1.84; 95% CI, 1.22-2.76). Postdialysis IP readmission was associated with suboptimal start (OR, 0.4; 95% CI, 0.17-0.93) and diabetes (OR, 2.14; 95% CI, 1.09-4.21). Black race (OR, 2.18; 95% CI, 1.44-3.31) was associated with postdialysis ED utilization.

DISCUSSION

The resource demands imposed by dialysis are a vexing problem for health system planners in VBC. Lack of timely nephrology referral and inadequate CKD care after referral are contributors to crash starts. Historically, health systems that were disconnected from upstream or downstream consequences of kidney failure relegated responsibility for ESKD care to outside dialysis organizations.¹⁹ Moreover, lacking clinical integration and financial alignment with dialysis organizations, health systems could do little to influence kidney care delivery. Traditional fee-for-service reimbursement provided perverse economic incentives to treat kidney failure and increase dialysis capacity as opposed to preventing the onset of ESKD. With the expansion of VBC and accountability for dialysis costs, health systems and payers must reexamine this approach. Our study provides important insights for health systems tackling the unacceptably high costs of dialysis.

Compared with US hospital visit data per 1000 population, crash starts have remarkably higher rates of IP and ED utilization, likely fueling costs.²⁰ Multivariable logistic regression showed that optimal starts were consistently and significantly associated with lower IP utilization, likely reflecting avoidance of IP dialysis initiation and subsequent CVC-related admissions. Crash starts may also represent a high-risk population with special needs. Diabetes, hypertension, and CHF are closely implicated with the progression of kidney failure and markers of cardiac prognosis and risk.⁴ Our multivariable logistic regression linked these comorbidities with higher IP admission, stressing the importance of comorbid disease management in CKD.

Suboptimal starts, representing 1 in 5 patients initiating dialysis, generated innovative findings. This group had the lowest predialysis utilization and expenditures. One explanation is that patients with suboptimal starts might somehow be healthier and/or less inclined to seek care. Indeed, this group had the lowest predialysis mean eGFR, suggesting some ability to tolerate worse kidney function. Paradoxically, suboptimal starts experienced a dramatic increase in postdialysis costs despite low IP and ED utilization. One possibility is suboptimal start costs rapidly “catch up” owing to vascular access procedures and other postponed interventions in the postdialysis period, albeit performed in the ambulatory setting. This novel group warrants further study, as they might not be detected by traditional utilization triggers and thus require special identification strategies.

Deficiencies in predialysis utilization seen in our study hint at system-level CKD care gaps. In our sophisticated health system, 45% of patients had no ambulatory predialysis nephrology care, highlighting an urgent need for earlier patient identification and referral to nephrologists.⁴ But referral alone may be insufficient, as the percentages of crash and optimal start patients seeing nephrologists were similar. The intensity and characteristics of predialysis CKD care likely make a difference. The optimal start group had significantly more nephrology visits than did the crash start group (mean, 3.24 vs 2.14; P < .001), perhaps reflecting greater predialysis planning. The low use of newer, game-changing CKD therapies such as SGLT2 inhibitors and GLP-1RAs is likely explained by the time period examined.²¹ Utilization of ACEIs or ARBs was low overall and unexpectedly highest for crash starts. Moreover, the crash start group had the highest frequency of eGFR testing at 3 and 6 months. Frequency of optimal CKD care (per our definition) was abysmally low overall, but highest for crash starts. These paradoxical findings do not diminish the importance of appropriate medications and monitoring, but rather they imply that other factors, including social determinants, strongly influence outcomes. Overall, our findings mirror the widespread failure to implement best practices and reinforce the need for holistic QI to address systematic health system barriers to CKD care.²¹

Our study reveals 3 key findings. First, health systems must ensure that patients are referred in a reliable manner for predialysis care. Second, the intensity of predialysis care must match individual patient needs, including differential care interventions targeting individual patient needs. Third, a new outcome category, suboptimal starts, is as costly as or more costly than crash starts. These findings reinforce the goals of the Advancing American Kidney Health Executive Order and Medicare KCC model.^8,9,22,23 PCP-specialist coordination, timely nephrology referral, appropriate intensity of CKD care, and adequate predialysis planning improve the likelihood of an optimal start.^8,9,22,23 Social determinants affecting CKD progression and access to care should also be addressed.

Limitations

Our study has some limitations. Our 12-month lookback limited more precise analysis of timing of nephrology visits and dialysis start categories. The mean number of visits was small for all groups, and we did not attempt to further correlate temporal patterns of visits with outcomes owing to limited sample size. Future investigations should examine utilization from the first-ever nephrology visit to dialysis start to better understand how frequency and longitudinal timing of visits are related. We were unable to identify patients with IP ESKD starts who died before discharge, so we excluded a subset of patients with likely high costs prior to death. PD is the lowest-cost treatment for ESKD, but our study had relatively few PD starts.⁴ With a larger PD cohort, the optimal start cost advantage may have been larger. We excluded preemptive transplant, which is classified as an optimal start by NQF but was out of scope for our QI project.⁹ Other studies have demonstrated the cost advantages of preemptive transplant compared with dialysis and we have no reason to believe that these economic assumptions are different for our population.^24,25 We lacked means to compare subgroups using a generalizable risk measurement (eg, Charlson Comorbidity Index). Milliman Advanced Risk Adjusters scores were partially available and are shown in eAppendix Table 3.

To convince health systems to invest in CKD care, the economic benefits of optimal start in VBC must be understood. We modeled different scenarios at CCF, showing aggregate savings ranging from $0.75 million with a 10% optimal start improvement to $3.8 million with a 50% improvement (eAppendix Table 4 [A]). Extrapolating our data to the incident US dialysis population, we estimated $165 million annual savings from a 10% national increase in optimal starts (eAppendix Table 4 [B]). Although these figures are speculative, they create a strong business case for health systems in VBC to pay attention to CKD. Additionally, our estimates do not include larger potential savings from delaying and preventing ESKD using highly efficacious new therapeutics such as SGLT2 inhibitors.^20,26 Health systems that do not realize the opportunity to reduce avoidable dialysis costs will find themselves at a competitive disadvantage in VBC.

CONCLUSIONS

In a large health system, crash and suboptimal dialysis starts were associated with increased costs. Promoting optimal starts is an important health system lever to improve VBC performance. Greater understanding of the factors influencing dialysis costs will help justify health system investments in better upstream CKD care to optimize coordination between PCPs and nephrologists, maximize utilization of kidney protective therapies, delay ESKD onset, and promote optimal dialysis starts. These study findings reinforce the national focus on optimal starts as a leading measure for high-value kidney care in the United States.

Acknowledgments

The authors thank Hunter Block-Beach, BS, Cleveland Clinic Community Care, and Ina Li, RN, MHA, Cleveland Clinic Glickman Urological and Kidney Institute, for providing invaluable data analyst support. Part of this work was presented in e-poster format at the virtual American Society of Nephrology Kidney Week meeting in November 2021.

Author Affiliations: Department of Kidney Medicine (LPW), Department of Internal Medicine (LPW, AG, JAH), Department of Biostatistics and Quantitative Health Sciences (JL), and Department of Gastroenterology, Hepatology, and Nutrition (MKR), Cleveland Clinic, Cleveland, OH.

Source of Funding: None.

Author Disclosures: Dr Wong was an employee of Cleveland Clinic at the time of writing this article. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (LPW, AG, MKR, JAH); acquisition of data (LPW, JAH); analysis and interpretation of data (LPW, JL, MKR, JAH); drafting of the manuscript (LPW, MKR, JAH); critical revision of the manuscript for important intellectual content (LPW, AG, JL, MKR, JAH); statistical analysis (JL, JAH); administrative, technical, or logistic support (LPW, AG, JAH); and supervision (LPW).

Address Correspondence to: Leslie P. Wong, MD, MBA, Intermountain Kidney Services and Nephrology, Intermountain Healthcare, 5169 S Cottonwood St, Ste 320, Murray, UT 84107. Email: leslie.wong1@imail.org.

REFERENCES

1. Health system participation in accountable care organizations (ACOs), 2016. Comparative Health System Performance Initiative data highlight #4. January 2019. Accessed February 28, 2022. https://www.ahrq.gov/sites/default/files/wysiwyg/chsp/aco-participation.pdf

2. Liao JM, Wang E, Isidro U, Navathe AS. The impact of simultaneous hospital participation in accountable care organizations and bundled payments on episode outcomes. Am J Med Qual. 2022;37(2):173-179. doi:10.1097/01.JMQ.0000754532.72567.c9

3. Shared Savings Program: program data. CMS. Accessed February 28, 2022. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/program-data

4. United States Renal Data System. 2021 USRDS Annual Data Report: epidemiology of kidney disease in the United States. National Institute of Diabetes and Digestive and Kidney Diseases. 2021. Accessed February 28, 2022. https://usrds-adr.niddk.nih.gov/2021

5. Bakre S, Hollingsworth JM, Yan PL, Lawton RJ, Hirth RA, Shahinian VB. Accountable care organizations and spending for patients undergoing long-term dialysis. Clin J Am Soc Nephrol. 2020;15(12):1777-1784. doi:10.2215/CJN.02150220

6. Nichols GA, Ustyugova A, Déruaz-Luyet A, O’Keeffe-Rosetti M, Brodovicz KG. Health care costs by type of expenditure across eGFR stages among patients with and without diabetes, cardiovascular disease, and heart failure. J Am Soc Nephrol. 2020;31(7):1594-1601. doi:10.1681/ASN.2019121308

7. Marrufo G, Colligan EM, Negrusa B, et al. Association of the Comprehensive End-Stage Renal Disease Care model with Medicare payments and quality of care for beneficiaries with end-stage renal disease. JAMA Intern Med. 2020;180(6):852-860. doi:10.1001/jamainternmed.2020.0562

8. Jain G, Weiner DE. Value-based care in nephrology: the Kidney Care Choices model and other reforms. Kidney360. 2021;2(10):1677-1683. doi:10.34067/KID.0004552021

9. Kidney Care Choices (KCC) model. CMS. Accessed February 28, 2022. https://innovation.cms.gov/innovation-models/kidney-care-choices-kcc-model

10. Mendu ML, Tummalapalli SL, Lentine KL, et al. Measuring quality in kidney care: an evaluation of existing quality metrics and approach to facilitating improvements in care delivery. J Am Soc Nephrol. 2020;31(3):602-614. doi:10.1681/ASN.2019090869

11. Brady BM, Ragavan MV, Simon M, Chertow GM, Milstein A. Exploring care attributes of nephrologists ranking favorably on measures of value. J Am Soc Nephrol. 2019;30(12):2464-2472. doi:10.1681/ASN.2019030219

12. Kalantar-Zadeh K, Kovesdy CP, Streja E, et al. Transition of care from pre-dialysis prelude to renal replacement therapy: the blueprints of emerging research in advanced chronic kidney disease. Nephrol Dial Transplant. 2017;32(suppl 2):ii91-ii98. doi:10.1093/ndt/gfw357

13. Liu HH, Zhao S. Savings opportunity from improved CKD care management. J Am Soc Nephrol. 2018;29(11):2612-2615. doi:10.1681/ASN.2017121276

14. Gillespie BW, Morgenstern H, Hedgeman E, et al. Nephrology care prior to end-stage renal disease and outcomes among new ESRD patients in the USA. Clin Kidney J. 2015;8(6):772-780. doi:10.1093/ckj/sfv103

15. Crooks PW, Thomas CO, Compton-Phillips A, et al. Clinical outcomes and healthcare use associated with optimal ESRD starts. Am J Manag Care. 2018;24(10):e305-e311.

16. Caro Martinez A, Olry de Labry Lima A, Muñoz Terol JM, et al. Optimal start in dialysis shows increased survival in patients with chronic kidney disease. PLoS One. 2019;14(7):e0219037. doi:10.1371/journal.pone.0219037

17. Optimal end stage renal disease (ESRD) starts. In: NQF-Endorsed Measures for Renal Conditions, 2015: Technical Report. National Quality Forum; 2015:168-172. Accessed February 28, 2022. https://www.qualityforum.org/Projects/n-r/Renal_Measures/Final_Report.aspx

18. Pravoverov LV, Zheng S, Parikh R, et al. Trends associated with large-scale expansion of peritoneal dialysis within an integrated care delivery model. JAMA Intern Med. 2019;179(11):1537-1542. doi:10.1001/jamainternmed.2019.3155

19. United States Renal Data System. 2016 USRDS Annual Data Report: epidemiology of kidney disease in the United States. National Institute of Diabetes and Digestive and Kidney Diseases. 2016. Accessed February 28, 2022. https://www.niddk.nih.gov/about-niddk/strategic-plans-reports/usrds/prior-data-reports/2016

20. State health facts: providers & service use: hospital utilization. Kaiser Family Foundation. Accessed July 25, 2022. https://www.kff.org/state-category/providers-service-use/hospital-utilization/

21. Tuttle KR, Wong L, St Peter W, et al; Diabetic Kidney Disease Collaborative Task Force. Moving from evidence to implementation of breakthrough therapies for diabetic kidney disease. Clin J Am Soc Nephrol. 2022;17(7):1092-1103. doi:10.2215/CJN.02980322

22. Lok CE, Huber TS, Lee T, et al; National Kidney Foundation. KDOQI clinical practice guideline for vascular access: 2019 update. Am J Kidney Dis. 2020;75(4 suppl 2):S1-S164. doi:10.1053/j.ajkd.2019.12.001

23. Executive Office of the President. Advancing American Kidney Health: executive order 13879 of July 10, 2019. Fed Regist. 2019;84(135):33817-33819. Accessed July 10, 2022. https://www.federalregister.gov/documents/2019/07/15/2019-15159/advancing-american-kidney-health

24. Axelrod DA, Schnitzler MA, Xiao H, et al. An economic assessment of contemporary kidney transplant practice. Am J Transplant. 2018;18(5):1168-1176. doi:10.1111/ajt.14702

25. Yang F, Liao M, Wang P, Yang Z, Liu Y. The cost-effectiveness of kidney replacement therapy modalities: a systematic review of full economic evaluations. Appl Health Econ Health Policy. 2021;19(2):163-180. doi:10.1007/s40258-020-00614-4

26. McEwan P, Bennett H, Khunti K, et al. Assessing the cost-effectiveness of sodium-glucose cotransporter-2 inhibitors in type 2 diabetes mellitus: a comprehensive economic evaluation using clinical trial and real-world evidence. Diabetes Obes Metab. 2020;22(12):2364-2374. doi:10.1111/dom.14162