Value-Based Care Interventions and Management of CKD Progression

ABSTRACT

Objectives: This study aimed to evaluate the effectiveness of value-based care (VBC) interventions in slowing the progression of chronic kidney disease (CKD), as measured by changes in estimated glomerular filtration rate (eGFR) over time.

Study Design: This retrospective cohort study used eGFR values from 2017 to 2023 to evaluate 623 patients with stage 3b or 4 CKD with routine eGFR testing who received VBC intervention. The focus was on comparing rates of eGFR decline before and after enrollment in the VBC.

Methods: Linear regression was used to model patient-specific trajectories of kidney function across time using eGFR, with the slope serving as an estimate for the rate of disease progression.Patients were grouped into cohorts by disease stage, and mixed-effects models were used to compare the rates of eGFR decline pre– and post VBC intervention.

Results: The rate of eGFR decline was slower across all CKD stages after enrollment compared with before, with a 77.2% reduction in the median rate of eGFR decline in stage 3b (P < .001) and 65.2% in stage 4 (P < .001). As a result of the reduced rates of disease progression, patients had higher median eGFR values at their latest month of eGFR measurement post enrollment compared with the eGFR values expected without a VBC intervention (stage 3b, 1.9 mL/min/1.73 m² higher; stage 4, 3.5 mL/min/1.73 m² higher).

Conclusions: Our findings indicate statistically significant differences in the rate of eGFR decline after enrollment in a VBC model, particularly for those in advanced CKD stages.

Am J Manag Care. 2025;31(10):e302-e307. https://doi.org/10.37765/ajmc.2025.89812

Takeaway Points

This study demonstrates that a value-based care (VBC) system can slow chronic kidney disease (CKD) progression, as evidenced by slower declines in estimated glomerular filtration rate (eGFR) among 623 patients.

• We evaluated the impact of intervention by a robust care team on patients with stage 3b or 4 CKD and various comorbidities. Further research with a larger and more diverse sample is needed to reinforce the hypothesis that VBC models can consistently slow disease progression across a range of CKD stages.
• Although providers use eGFR results to assess kidney function, no national programs exist to incentivize providers to delay disease progression by monitoring rates of eGFR decline.

Chronic kidney disease (CKD) affects 14% of adults in the US,¹ yet up to 90% remain unaware of their condition.² As a result, CKD often goes undiagnosed and can progress to kidney failure (end-stage kidney disease; ESKD), requiring dialysis or kidney transplantation. CKD is particularly prevalent among patients with diabetes and hypertension³ and is linked to an increased risk of congestive heart failure (CHF).⁴ CKD is a leading cause of global mortality⁵ and has resulted in more than $130 billion in annual Medicare spending in the US.⁶ Early detection and management of CKD, along with associated chronic medical conditions such as hypertension, diabetes, and CHF, are essential for timely interventions that can slow CKD progression, reduce the risk of kidney failure, and enhance patient outcomes.⁷

Value-based care (VBC) models prioritize quality clinical interventions more than fee-for-service (FFS) models, proactively providing care coordination and emphasizing engagement by high-risk patients. In the US, there is growing interest in VBC for patients with kidney disease, with federal programs offering strong incentive models for providers to effectively manage their care.⁸ This patient-centric approach allows for early monitoring and treatment of CKD, improving patient outcomes through patient education, shared decision-making, and complex care management.

In an in-depth review of renal function trajectory, Rosansky noted variability in disease progression among patients with similar CKD stages and advised that renal function trajectory can inform comprehensive care plans and preparation for renal replacement therapy (RRT).⁹ The value-based kidney care organization in the current study employs an integrated care delivery model with high-touch care interventions (Table 1) and predictive analytics to identify the highest-risk patients in order to prioritize engagement in clinical interventions. For example, a patient recently hospitalized for CHF exacerbation may be flagged as a high priority and may be enrolled to address barriers to care and develop an individualized treatment plan. This approach facilitates early disease detection and coordinated care to slow CKD progression through lifestyle modifications such as medication management, smoking cessation, and dietary modifications.

Estimated glomerular filtration rate (eGFR) is a key indicator of kidney function, and a decrease in eGFR over time reflects worsening kidney function. The overall rate of eGFR decline is a critical estimate of the rate of CKD progression. We evaluated the effectiveness of an intentionally crafted care delivery model by analyzing eGFR trends before and after enrollment in the program, assessing the impact on rates of CKD progression.

METHODS

Patient Population and Data

The study cohort consisted of patients 18 years or older who had commercial, Medicare Advantage, or Medicare FFS health insurance plans; were diagnosed with CKD stages 2 through 5; and enrolled in the program between May 2020 and November 2023. eGFR results between 2017 and 2023 were calculated for patients enrolled in the VBC program using the clinically validated, unbiased, race-free CKD Epidemiology Collaboration equation, the recommended method by the National Kidney Foundation (NKF).^10,11 CKD stage was determined by eGFR value at enrollment (ie, baseline) using NKF-established average ranges.¹² The study team operates within the health care sector, partnering with insurance health plans and providers (primary care providers [PCPs] and nephrologists) nationwide with a focus on improving patient outcomes (eg, stabilizing chronic conditions) and reducing avoidable costs (eg, preventable hospitalizations) for individuals with kidney disease. Medical claims data and enrollment records were linked with patient demographics and used to identify plan-eligible patients and comorbidities. Dialysis use and the diagnosis of comorbid conditions known to influence CKD progression (ie, diabetes, hypertension, CHF) were identified from medical claims.¹³

A power analysis was conducted to estimate the sample sizes necessary to generate robust conclusions from the study. Requiring 80% power and a significance level of α = 0.05, we estimated that the VBC would have a small effect (Cohen d = 0.2) on earlier CKD stages (2, 3a) and a moderate effect (d = 0.5) on later stages (3b, 4, 5). As a result, the study required 199 patients with stage 2 or 3a CKD and 34 patients with stage 3b, 4, or 5 CKD.

Rigorous inclusion and exclusion criteria were established to ensure that data associated with the study population were clearly defined and indicative of chronic disease progression. As a result of requiring enough eGFR values to model chronic trends over time, the final patient cohort represents a population with more routine eGFR testing. Patients who transitioned to ESKD before enrollment were excluded entirely from the analysis. eGFR results captured after a kidney transplant or transition to ESKD were excluded, as were eGFR results within 3 months of an acute kidney injury diagnosis or dialysis treatment and those obtained while a patient was in hospice care. If multiple eGFR values existed in a single month, the lowest (ie, worst) eGFR value was used so that a single eGFR was an indicator of each patient’s minimal kidney function in the month. eGFR was analyzed across all available months pre- and post enrollment, with the month of enrollment considered neutral, such that eGFR data in this month were not used for either the pre- or postenrollment period. In addition, the duration of eGFR data for each patient was matched between time periods so that the pre- and postenrollment periods for each patient utilized the same time span of data: The time period with a longer duration of data available was truncated to match the shorter duration.

Patients with extreme variability in their eGFR surrounding enrollment were excluded if the mean absolute difference (MAD) between the last 4 eGFR values preenrollment and the first 4 eGFR values post enrollment was greater than 10. To ensure adequate longitudinal data to accurately estimate the slope of eGFR decline based on multiple data points,¹⁴ patients were required to have at least 4 distinct months with eGFR measurements across a 12-month or longer period preenrollment and a 12-month or longer period post enrollment; one exception to this requirement was if a patient transitioned to ESKD within 12 months post enrollment. Patients were also excluded if they had no recorded eGFR measurements within the 6 months pre- and 6 months post enrollment or if the MAD between their last preenrollment and first postenrollment eGFR values was greater than 10 mL/min/1.73 m². Full inclusion and exclusion characteristics, descriptions, reasoning for each criterion implemented, and relevant medical codes are displayed in eAppendix Tables 1 and 2 (eAppendix available at ajmc.com).

Outcomes

To assess the impact of a VBC program on disease progression, we modeled patient-specific eGFR trajectories using linear regression and separated patients into CKD stage–specific cohorts. We calculated individual rates of eGFR decline per year before and after enrollment, then aggregated these rates to compare the observed and expected (based on preenrollment trajectories) rate of eGFR decline per CKD stage cohort. The expected eGFR value was then calculated for the latest eGFR result month in the postenrollment period for each patient and compared with their actual eGFR result in that month. Together, these metrics can provide an estimate of the total impact of enrollment in a VBC program on disease progression within each CKD stage cohort.

Modeling Kidney Function Trajectories

Linear regression modeling was used to estimate the rate of chronic decline in kidney function for each patient during each time period. These models were fit to the time-series eGFR data for each patient, with the slope representing the rate of decline in kidney function over time, expressed in mL/min/1.73 m² per year. Preenrollment and postenrollment periods were fitted separately by the linear regression models to obtain their rate of decline per year. The slopes from the individual patient models were aggregated by CKD stage for each time period, with the median slope describing the overall rate of CKD progression for patients in each cohort.

A comparison of disease progression with and without the VBC intervention can be made by interpreting the preenrollment rates of eGFR decline as the expected rate of disease progression and contrasting this with the observed postenrollment rates to calculate potential differences in eGFR values over time. The expected eGFR value for each patient is calculated by adding the first eGFR value post enrollment (baseline eGFR) to the product of the cohort-median preenrollment rate of eGFR decline (slope) and duration of months between the first and latest eGFR post enrollment. These expected eGFR values and the latest observed eGFR values post enrollment for each patient were aggregated by CKD stage cohort and compared.

A paired, 1-sided Wilcoxon signed rank test was used to test for significant differences between the 2 closest eGFRs to the time of enrollment: the latest eGFR preenrollment and the first eGFR post enrollment. Mixed-effects modeling was employed to test for a significant difference in the rate of eGFR decline between the pre- and postenrollment periods for each CKD stage cohort. This model allowed for the inclusion of both fixed effects (time) and random effects (individual patient variability) and accommodated the nonparametric data.¹⁵ Lastly, Cohen d was used to estimate the effect size of the observed differences between time periods.

RESULTS

A total of 623 patients were included in this study, and baseline characteristics are summarized in Table 2. The number of patients meeting the inclusion and exclusion criteria for CKD stages 2, 3a, or 5 were fewer than the minimum sample size requirements (CKD stage 2, n = 33; CKD stage 3a, n = 99; CKD stage 5, n = 30), so the analysis was performed for those with CKD stage 3b (n = 310) or CKD stage 4 (n = 313) only. The study participants consisted largely of older adults diagnosed with comorbidities and a slightly higher proportion of female patients. Patients typically had 1.5 years of eGFR data during each time period, with the earliest and latest eGFR results occurring at nearly 2 years pre- and post enrollment (eAppendix Table 3).

The median eGFR values just prior to and after enrollment were similar for both stages, but the underlying distributions showed significantly lower eGFRs post enrollment for CKD stage 4 (CKD stage 3b, P = .089; CKD stage 4, P = .037) (Table 3¹²). Due to this observation, the expected eGFRs at the latest time point post enrollment (based on preenrollment rates of decline) were calculated starting from each patient’s first eGFR post enrollment rather than from an estimated eGFR at the time of enrollment (model intercept at time 0).

Rates of eGFR decline slowed after enrollment for patients in both CKD stages (Figure and Table 4). In patients with CKD stage 3b, the median rate of decline slowed from –1.45 to –0.33 mL/min/1.73 m² per year (P < .001; d = 0.42 [95% CI, 0.26-0.58]), corresponding to a 77.2% reduction in the rate of CKD progression. In patients with CKD stage 4, the median rate of decline slowed from –2.90 to –1.01 mL/min/1.73 m² per year (P < .001; d = 0.67 [95% CI, 0.51-0.83]), resulting in a 65.2% reduction in the rate of progression. The median eGFR was 1.9 mL/min/1.73 m² higher than expected for patients with CKD stage 3b (observed vs expected, 33.5 vs 31.6 mL/min/1.73 m²) and 3.5 mL/min/1.73 m² higher for patients with CKD stage 4 (observed vs expected, 23.0 vs 19.5 mL/min/1.73 m²). Overall, patients had notably higher eGFR values at their latest month of eGFR measurement post enrollment compared with the eGFR values expected in that month based on their preenrollment eGFR trajectory. These results indicate a significant slowing of kidney function decline after enrollment in the VBC program.

DISCUSSION

These results suggest that enrollment in a VBC program delivering high-touch care to patients with routine eGFR testing was associated with slowing of CKD progression. The rate of eGFR decline was more rapid in the period prior to enrollment, and a statistically significant reduction in the rate of CKD progression was noted after enrollment. We observed a stronger effect of enrollment on the delay of progression in patients with CKD stage 4 (moderate effect size) than CKD stage 3b (small effect size), suggesting a more robust impact of VBC interventions in late-stage kidney disease compared with earlier stages. The potential to slow CKD stage 4 progression is especially critical as it delays the need for dialysis and allows more time to prepare for RRT, ideally preventing a crash dialysis start and increasing quality of life. Using their latest eGFR measure post enrollment, we saw that patients with CKD stage 3b had progression of 76.7% through stage 3b compared with the expected 89.3% progression (CKD stage 3b range, 30 ≤ eGFR < 45 mL/min/1.73 m²). Similarly, patients with CKD stage 4 had progression through only 46.7% of stage 4 compared with the expected 70.0% (CKD stage 4 range, 15 ≤ eGFR < 30 mL/min/1.73 m²). Although the instantaneous rate of CKD progression across time will be variable for most patients, if the general trend of eGFR decline can be slowed by similar magnitudes to what we have observed in this study, kidney failure may be delayed by months to years for some patients.

Our study contributes to the growing body of research for slowing kidney disease by employing a more diverse high-touch care team (eAppendix Table 4). Previous research, such as work from Bayliss et al¹⁶ on patients with CKD stage 3 with associated comorbidities, showed a slower rate of eGFR decline in patients referred to multidisciplinary care compared with general nephrology care. Chen et al¹⁷ connected patients with CKD stages 3 to 5 to high-touch multidisciplinary care, noting a slower decline in eGFR in those with CKD stages 4 or 5 compared with patients receiving standard PCP-guided care management, although no significant difference in renal function between the multidisciplinary care group and PCP-guided group was observed in early-stage CKD (stage 3). Consistent with the findings of Imamura et al,¹⁸ which highlighted substantial improvement in eGFR decline following multidisciplinary care in patients with CKD regardless of diabetes status, our study also observed a profound reduction in rate of progression. By engaging a more robust nephrology-guided care team, the current study findings demonstrate the profound impact VBC interventions may have on managing kidney disease (Table 4).

A critical aspect of patient care management is medication reconciliation and goal-directed medical therapy. However, medications appropriate for many patients with CKD (eg, sodium-glucose cotransporter 2 [SGLT2] inhibitors and renin-angiotensin-aldosterone system [RAAS] inhibitors) can cause an acute decrease in eGFR when use is initiated, followed by an attenuation of the slope of eGFR decline with chronic use.^19,20 When included in our statistical model as cofactors, initiation of SGLT2 inhibitor or RAAS inhibitor medications did not modulate the significant effect of the intervention on the change in rate of eGFR (P < .001), confirming that our study results were not biased by initiation of these medications.

CMS offers financial incentives to dialysis facilities, nephrologists, and other health care providers to manage and delay progression to advanced stages of kidney disease (CKD stages 4-5 and ESKD) and delay RRT. These incentives are designed to align with health outcomes and utilization for the patient, as well as national standards, and include bonus incentives for each successful kidney transplant.⁸ Today, there are no explicit quality metrics targeting the proactive delay of disease progression by monitoring the rate of eGFR decline; existing measures instead focus on the point of transition to ESKD and dialysis. Although the trajectory of disease progression varies widely and typically fluctuates across time, tracking changes in eGFR over time may help providers to understand patient prognoses and provide more timely intervention. Ideally, if a change in rate of eGFR decline is observed in moderate-stage CKD, actions can be taken to preserve residual kidney function and attenuate the decline. In late-stage CKD, precipitous declines can be identified early, allowing for adequate preparation for RRT and transplant education, preventing inpatient dialysis crash starts. Adding proactive, eGFR-decline reporting metrics with existing CMS incentives may help to support the broader goals of optimizing patient health span and reducing avoidable health care use.

Limitations

A fundamental limitation of the study is the comparison of pre- and postenrollment time periods, as opposed to a matched comparison between treated and nontreated patients. The choice in study design leading to this limitation was due to poor longitudinal data availability for patients who were eligible for the VBC program but never enrolled. eGFR data are often collected in clinical settings, and access to these data is limited to results provided by the insurance health plans and third-party laboratory vendors. A future study with sufficient nonenrolled patient data over time should compare VBC-enrolled patients with a representative nontreated cohort using a patient-matching method such as propensity score matching.²¹

We found that most patients who qualified for the study were those who had 1 or more comorbidities, were older, and were in a later CKD stage. We hypothesize that the patients more likely to get regular laboratory testing at the recommended frequency²² are those who have more complex health conditions and visit their health care provider more often, causing potential selection bias.

CKD progression is typically slow, estimated as taking 7 to 25 years to advance from stage 3a to 5.²³ This study analyzed eGFR decline over an observation period of 20 months, which may make it difficult to detect significant changes, especially for disease stage cohorts with small sample sizes and in earlier stages where patients spend more time and have a slower rate of kidney decline. For example, Ku et al²³ estimated that the time spent in stage CKD 3a is typically approximately 7.9 years compared with 0.8 years in stage 5. Future studies of VBC programs should include longer follow-up periods to further advise clinicians on the impact of specific interventions on disease progression throughout all CKD stages.

Additionally, eGFR decline does not occur at a constant rate in most patients, even within the same stage. On average, the rate of decline accelerates as the disease progresses, but patients may have periods of steep decline or plateaus in kidney function.²³ Although slopes generated by linear regression models are easy for clinicians to understand and compare across time when expressed as eGFR decline per year, alternative data models could benefit from quantifying the nonlinear progression of kidney decline over time and produce more precise results by controlling for patient characteristics as covariates.²⁴ These and other factors complicate efforts to accurately assess the effectiveness of interventions in delaying disease progression.

CONCLUSIONS

We found that patients with stage 3b or 4 CKD enrolled in the VBC program had slower rates of disease progression than expected if the patients had not enrolled. Our findings demonstrate that enrollment in a VBC organization targeting kidney disease may be successful in delaying renal function.

Acknowledgments

The authors thank Mohammadreza Faraji, PhD, and Matthew Temba.

Author Affiliations: Strive Health (MF, JM, TC, ES, FM), Denver, CO.

Source of Funding: None.

Author Disclosures: All authors are full-time employees of Strive Health, and Dr Modarai has vested and unvested stock in Strive Health.

Authorship Information: Concept and design (MF, TC, ES, FM); acquisition of data (MF); analysis and interpretation of data (MF, JM, FM); drafting of the manuscript (MF, JM, TC, ES, FM); critical revision of the manuscript for important intellectual content (MF, JM, TC, ES); statistical analysis (MF, JM); administrative, technical, or logistic support (ES); and supervision (TC, ES, FM).

Address Correspondence to: Emily Simon, BS, Strive Health, 1125 17th St #1000, Denver, CO 80202. Email: esimon@strivehealth.com.

REFERENCES

1. United States Renal Data System. 2023 USRDS Annual Data Report: Epidemiology of Kidney Disease in the United States. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases; 2023. Accessed August 23, 2024. https://usrds-adr.niddk.nih.gov/2023

2. Chronic kidney disease in the United States, 2023. CDC. Updated May 15, 2024. Accessed August 23, 2024. https://www.cdc.gov/kidney-disease/php/data-research/index.html

3. Chen TK, Knicely DH, Grams ME. Chronic kidney disease diagnosis and management: a review. JAMA. 2019;322(13):1294-1304. doi:10.1001/jama.2019.14745

4. Vijay K, Neuen BL, Lerma EV. Heart failure in patients with diabetes and chronic kidney disease: challenges and opportunities. Cardiorenal Med. 2022;12(1):1-10. doi:10.1159/000520909

5. Kovesdy CP. Epidemiology of chronic kidney disease: an update. Kidney Int Suppl (2011). 2022;12(1):7-11. doi:10.1016/j.kisu.2021.11.003

6. Federal investment. National Kidney Foundation. Accessed August 23, 2024. https://www.kidney.org/federal-investment

7. Pontillo C, Jacobs L, Staessen JA, et al. A urinary proteome-based classifier for the early detection of decline in glomerular filtration. Nephrol Dial Transplant. 2016;32(9):1510-1516. doi:10.1093/ndt/gfw239

8. Kidney Care Choices (KCC) model. CMS. Accessed August 23, 2024. https://www.cms.gov/priorities/innovation/innovation-models/kidney-care-choices-kcc-model

9. Rosansky SJ. Renal function trajectory is more important than chronic kidney disease stage for managing patients with chronic kidney disease. Am J Nephrol. 2012;36(1):1-10. doi:10.1159/000339327

10. Inker LA, Eneanya N, Coresh J, et al; Chronic Kidney Disease Epidemiology Collaboration. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med. 2021;385(19):1737-1749. doi:10.1056/NEJMoa2102953

11. Delgado C, Baweja M, Crews DC, et al. A unifying approach for GFR estimation: recommendations of the NKF-ASN task force on reassessing the inclusion of race in diagnosing kidney disease. J Am Soc Nephrol. 2021;32(12):2994-3015. doi:10.1681/ASN.2021070988

12. Stages of chronic kidney disease (CKD). National Kidney Foundation. Accessed August 23, 2024. https://www.kidney.org/kidney-topics/stages-chronic-kidney-disease-ckd

13. MacRae C, Mercer SW, Guthrie B, Henderson D. Comorbidity in chronic kidney disease: a large cross-sectional study of prevalence in Scottish primary care. Br J Gen Pract. 2021;71(704):e243-e249. doi:10.3399/bjgp20X714125

14. Nojima J, Meguro S, Ohkawa N, Furukoshi M, Kawai T, Itoh H. One-year eGFR decline rate is a good predictor of prognosis of renal failure in patients with type 2 diabetes. Proc Jpn Acad Ser B Phys Biol Sci. 2017;93(9):746-754. doi:10.2183/pjab.93.046

15. Detry M, Ma Y. Analyzing repeated measurements using mixed models. JAMA. 2016;315(4):407-408. doi:10.1001/jama.2015.19394

16. Bayliss EA, Bhardwaja B, Ross C, Beck A, Lanese DM. Multidisciplinary care team may slow the rate of decline in renal function. Clin J Am Soc Nephrol. 2011;6(4):704-710. doi:10.2215/CJN.06610810

17. Chen YR, Yang Y, Wang SC, et al. Effectiveness of multidisciplinary care for chronic kidney disease in Taiwan: a 3-year prospective cohort study. Nephrol Dial Transplant. 2012;28(3):671-682. doi:10.1093/ndt/gfs469

18. Imamura Y, Takahashi Y, Hayashi T, et al. Usefulness of multidisciplinary care to prevent worsening renal function in chronic kidney disease. Clin Exp Nephrol. 2018;23(4):484-492. doi:10.1007/s10157-018-1658-z

19. Jongs N, Chertow GM, Greene T, et al; DAPA-CKD Trial Committees and Investigators; Members of the DAPA-CKD Trial Committees and Investigators. Correlates and consequences of an acute change in eGFR in response to the SGLT2 inhibitor dapagliflozin in patients with CKD. J Am Soc Nephrol. 2022;33(11):2094-2107. doi:10.1681/ASN.2022030306

20. Weir MR. Acute fall in glomerular filtration rate with renin-angiotensin system inhibition: a biomeasure of therapeutic success? Kidney Int. 2011;80(3):235-237. doi:10.1038/ki.2011.132

21. Futoma J, Sendak M, Cameron CB, Heller K. Scalable joint modeling of longitudinal and point process data for disease trajectory prediction and improving management of chronic kidney disease. Uncertain Artif Intell. 2016. Accessed September 25, 2024. https://auai.org/uai2016/proceedings/papers/160.pdf

22. Hirano K, Kobayashi D, Kohtani N, et al. Optimal follow-up intervals for different stages of chronic kidney disease: a prospective observational study. Clin Exp Nephrol. 2019;23(5):613-620. doi:10.1007/s10157-018-01684-4

23. Ku E, Johansen KL, McCulloch CE. Time-centered approach to understanding risk factors for the progression of CKD. Clin J Am Soc Nephrol. 2018;13(5):693-701. doi:10.2215/CJN.10360917

24. Li L, Astor BC, Lewis J, et al. Longitudinal progression trajectory of GFR among patients with CKD. Am J Kidney Dis. 2012;59(4):504-512. doi:10.1053/j.ajkd.2011.12.009