The cost implications of first anniversary renal function after living, standard criteria deceased and expanded criteria deceased donor kidney transplantation

Abstract

Objectives:

To quantify relationships of post-transplant renal function with healthcare costs after kidney transplantation.

Methods:

Clinical and billing records for Medicare-insured kidney transplant recipients (1995–2003) were drawn from the US Renal Data System. Estimated glomerular filtration rate (eGFR) at 1-year post-transplant was computed with the abbreviated Modification of Diet in Renal Disease equation. Associations of eGFR with total Medicare payments in the second and third post-transplant years were examined by multivariate non-linear regression with spline forms. Adjustment covariates were drawn from the survival prediction model developed by the UNOS Kidney Allocation Review Committee.

Results:

The sample comprised 7103 living donor (LD), 22,110 standard criteria deceased (SCD), and 2594 expanded criteria deceased (ECD) donor transplant recipients. Regardless of donor type, lower 1-year eGFR was associated with significantly increased expenditures during the second and the third years post-transplant. Marginal costs began to increase as eGFR fell below 45 mL/min/1.73 m² and rose in an accelerating manner. Compared to a reference eGFR of 75 mL/min/1.73 m², 1-year eGFR of 20 ml/min/1.73 m² in SCD recipients was associated with ∼ $17,500 and $18,200 higher adjusted payments in the second and third post-transplant years, respectively. Patterns were similar among recipients of LD and ECD transplants.

Limitations:

The study sample was limited to Medicare beneficiaries who survived with allograft function to the first transplant anniversary, which may limit generalizability of the findings. eGFR is a surrogate measure of renal function. The design is retrospective and changes in post-transplant management may alter long-term cost implication of renal function.

Conclusions:

Decreased renal function is significantly associated with higher healthcare expenditures following kidney transplantation. Post-transplant eGFR may be a useful metric for discriminating the economic impact of care strategies that differentially affect renal function.

Keywords::

• Billing claims
• Glomerular filtration rate
• Healthcare costs
• Kidney transplantation
• Medicare

Introduction

Chronic kidney disease (CKD) is recognized as a source of substantial morbidity and economic burdens in the general, non-transplanted population. Epidemiologic studies have demonstrated associations between the degree of kidney dysfunction and an increased risk of hospitalization and healthcare resource utilization1,2. Data from a large Healthcare Maintenance Organization (HMO) demonstrated that patients with CKD stages 2–4 had 1.9–2.5-times more prescriptions, 1.3–1.9-times more outpatient visits, and 1.8–4.2-times more inpatient stays than age and gender-matched controls without CKD2. Notably, costs were similar across these functional stages within the CKD sample. In a recent analysis of privately insured patients with native autosomal dominant polycystic kidney disease, total charges did not differ significantly among patients with estimated glomerular filtration rate (eGFR) > 30 ml/min/1.73 m² after adjustment for age and gender, but rose precipitously when eGFR fell below 30 ml/min/1.73 m² 3.

The natural history of renal function after kidney transplantation is generally characterized by gradual but steady decline4, although the rate of decline varies across individuals based on multiple factors including donor quality, recipient comorbidities, immunologic risk, and immunosuppressive regimen. Allograft failure is an expensive event, incurring ∼ $100,000 in excess Medicare payments over the cost of expected post-transplant care in the subsequent year after failure5. We recently found that moderately reduced allograft function (eGFR 30–44 mL/min/1.73 m²) was associated with an average $1555 (p < 0.0001) increase in marginal healthcare costs in the second year post-transplant compared to eGFR ≥ 60 mL/min/1.73 m², and that total second year costs increased by an average of $5352 (p < 0.0001) among those with eGFR 15–30 mL/min/1.73 m² (year 2003 dollars)6. However, the cost implications of renal dysfunction after kidney transplant have not been precisely quantified across a spectrum of function levels and donor types in large, contemporary samples.

To advance the understanding of the impact of kidney graft function on healthcare costs in transplant recipients, we examined national registry data for a recent cohort of transplant recipients in the US. The primary goal was to determine the relationship of kidney function at the first transplant anniversary, as measured by eGFR, with healthcare costs during the secondary and third years after transplant. A unique methodological feature of this study is use of non-linear spline regression to flexibly quantify cost relationships across incremental changes in eGFR. Relationships were examined separately according to living donor (LD), standard criteria deceased (SCD), and expanded criteria deceased (ECD) donor source.

Materials and methods

Data sources

Study data were drawn from integrated records of the US Renal Data System (USRDS). The USRDS is a joint effort of the National Institute of Diabetes and Digestive and Kidney Disease (NIDDK) and the Centers for Medicare and Medicaid Services (CMS) that tracks many descriptive elements for all patients in the US ESRD program. USRDS databases integrate information from the Organ Procurement and Transplantation Network (OPTN), CMS, and Medicare billing claims records. These elements are linked with a unique encrypted patient identifier, permitting investigators to combine patient-specific information from multiple tables without revealing patient identity7. This study was conducted in accordance with the Health Insurance Portability and Accountability Act of 1996, and all standards regarding the security and privacy of an individual’s health information were maintained.

Study samples

The sample was drawn from recipients of single-organ kidney transplants in the US in 1995–2003 with Medicare as the primary payer at transplantation. We identified Medicare beneficiaries as those with Medicare ‘primary payer’ status indicated in the ‘Payer History’ file of the USRDS at the time of transplant. To ensure complete Medicare billing, we also required that the Medicare payment for the initial transplant hospitalization was at least $15,000, as previously described8. Intervals of Medicare insurance following transplant were identified from USRDS Payer History records. ECD transplants were defined according to UNOS/OPTN criteria as allografts from deceased donors ≥ 60 years of age or those from donors aged 50–59 years old with at least two of the following: history of hypertension, terminal serum creatinine > 1.5 mg/dL, or cerebrovascular cause of death9. Other transplants from deceased donors were defined as SCD.

Patients who died or experienced graft failure prior to the first transplant anniversary were excluded. Patients were also excluded if a required data element for calculation of eGFR at 1-year post-transplant (serum creatinine level, recipient age, gender, or race) was not recorded in the database. Finally, analytic samples for the second-year and third-year economic analyses were limited to patients with Medicare at the start of a given period who sustained coverage to the period end, or died or experienced graft failure within the period.

Predictor variable definitions

Clinical variables defined at transplant and at the first transplant anniversary were ascertained from the UNOS Transplant Recipient Registration and Recipient Follow-up Forms, respectively. The main predictor variable of interest was renal function at the first post-transplant anniversary, as quantified by eGFR. eGFR was computed according to the abbreviated Modification of Diet in Renal Disease (MDRD) equation as: eGFR (ml/min/1.73 m²) = 186 × (Serum Creatinine mg/dL)^−1.154 × Age^−0.203 × (1.212, if African-American) × (0.742, if Female)10. The abbreviated Modification of Diet in Renal Disease equation has superior performance for prediction of measured GFR among renal transplant patients when compared with the Nankivell and Cockcroft-Gault formulas11. We defined acute rejection according to center reporting that an acute rejection event occurred. Immunosuppression records were used to sub-classify acute rejection as antibody-treated (Ab-treated) or non-Ab-treated as a measure of acute rejection severity, per our previous methods6,12,13. Ab-treated acute rejection was defined by administration of polyclonal antibodies such as anti-thymocyte globulin or anti-lymphocyte globulin, or monocloncal antibodies such as OKT3, alemtuzumab, or rituximab for the indicated purpose of treating acute rejection. Patients with any Ab-treated acute rejection event in a period were classified as having Ab-treated acute rejection in that period, as the first level of classification. Patients with other indications of acute rejection in a period who did not meet criteria for Ab-treated acute rejection were classified as having non-Ab-treated acute rejection in the given period.

Outcome variable definitions

The primary economic measure was actual payments for all healthcare services made by Medicare. Payments were evaluated at 1-year intervals during the second and third years post-transplant. Cost ascertainment was limited to the third transplant anniversary, as Medicare transplant benefits expire at 3 years except in the cases of people aged ≥ 65 years or with certain disabilities. Patient costs were included in analysis of an interval if: (1) the recorded Medicare eligibility extended continuously from the beginning to the end of the period, or (2) Medicare eligibility ended in an interval because of death or graft loss, as per our previous methods6,12. Monetary figures were adjusted to the prices in the year 2003 Medical Care Component of the Consumer Price Index14.

Statistical analysis

Data management and analysis were performed with SAS for Windows software, version 9.2 (SAS Institute Inc., Cary, NC). Continuous data were summarized as means and standard deviations, and categorical data were summarized as proportions. Separate ordinary least squares (OLS) regression models were developed for total costs during the second and third years post-transplant. A flexible non-linear technique known as smoothed natural cubic splines was used to optimize model fit. Spline fitting of curves was first described by Schoenberg15 in 1946. The smoothed natural splines employed here are cubic, or third power, polynomial expansions of an independent variable of interest in a regression equation16, in this case eGFR. The reference function level for eGFR-related risk was set at eGFR 75 ml/min/1.73 m². Basis knots, which define the form of the polynomial expansion, where chosen to follow the K/DOQI ‘Stages of Chronic Kidney Disease’ starting at 15 mL/min/1.73 m² with additional knots at intervals of 15 mL/min/1.73 m² through 150 mL/min/1.73 m². An additional knot at 22.5 mL/min/1.73 m² was found to significantly increase fit in and was included in all models. Equations for the spline functions in SAS are provided in the Appendix. The joint significance of 1-year eGFR in the prediction of costs was assessed using an F-test of the null hypothesis that the spline parameters do not contribute significantly to the prediction of costs.

The models in this study used the structure and covariates developed by the UNOS Kidney Allocation Committee to predict survival after kidney transplantation17. Demographic and clinical characteristics known at the time of transplant were included in exact accordance with the UNOS models, with the exception of shared organ status, which was not present in the USRDS database available for public use. Recipient and donor race are omitted from the UNOS models but were included here (Appendix). We also included first anniversary eGFR and acute rejection. No statistical variable selection was performed, such that the content of all regression models was determined prior to analysis.

Results

Demographic and clinical characteristics

There were 126,073 kidney transplants recorded in the USRDS database during the study period. Of these, 14,061 were excluded due to graft loss before the first post-transplant anniversary; 22,013 were excluded due to missing 12-month OPTN follow-up record; and 2424 were excluded due to missing one or more of the data elements required to calculate eGFR at 1-year post-transplant. Of the remaining 87,575 patients, 31,807 had Medicare-primary insurance at transplantation.

The distribution of recipient, donor, and transplant characteristics varied across donor types (Table 1). Overall, the majority of the study sample received SCD organs (n = 22,110); 7103 received LD organs, and 2594 received ECD organs. Recipients of LD allografts were younger, more commonly white race, and were less likely to have diabetes or high levels of sensitization compared to deceased donor recipients. ECD recipients tended to be the oldest group and to have the highest frequency of diabetes. Mean eGFR at 12-months post-transplant was similar among recipients of LD (56.5 mL/min/1.73 m²) and SCD (53.4 mL/min/1.73 m²) transplants, while mean eGFR at 12-months post-transplant was substantially lower in ECD recipients (37.1 mL/min/1.73 m²).

Table 1.  Characteristics of Medicare-insured US kidney transplant recipients (1995–2003) who survived with graft function to the first transplant anniversary, by donor type.

	LD	SCD	ECD
Number of subjects	7103	22,110	2594
Candidate/recipient variables
Mean eGFR at 12 months post-transplant	56.5 ± 26.1	53.4 ± 37.1	37.1 ± 15.8
Recipient age (years)	42.8 ± 16.6	46.2 ± 14.0	52.2 ± 12.0
Male	56.7%	60.6%	60.4%
Pre-emptive transplant	49.4%	5.9%	4.6%
Years of dialysis (mean)	3.1 ± 1.23	3.8 ± 1.52	4.2 ± 1.43
Body mass index (kg/m^2)	25.9 ± 5.6	26.2 ± 5.7	27.2 ± 5.4
Body mass index missing	25.5%	28.7%	29.4%
Cause of ESRD
Glomerulonephritis or diabetes	43.5%	41.2	42.7%
Hypertension	17.1%	20.9%	24.1%
Polycystic kidney disease	4.6%	7.0%	8.4%
Other	34.8%	30.9%	26.7%
Diabetes	25.1%	26.5%	28.6%
Previous transplant	13.0%	16.2%	10.9%
Peak panel reactive antibody level
< 10	75.6%	66.0%	69.8%
10–79	16.9%	23.2%	22.4%
≥ 80	3.5%	8.4%	5.6%
Missing	4.0%	2.0%	2.2%
HLA mismatches
Zero A, B, and DR	12.5%	13.6%	10.9%
Zero DR	26.7%	23.3%	22.4%
Serum albumin at listing (g/dl)
Serum albumin	3.8 ± 0.7	3.8 ± 0.7	3.8 ± 0.7
Missing	78.5%	88.9%	91.5%
Race
White	75.5%	63.8%	65.0%
African-American	19.5%	30.5%	30.0%
Other	5.0%	5.7%	2.8%
Donor variables
Age (in years)	38.3 ± 14.8	31.4 ± 11.0	61.2 ± 6.5
Missing	0.0%	12.2%	0%
Hypertension	0%	12.9%	57.3%
Weight (kg)	79.1 ± 23.2	71.3 ± 16.2	74.4 ± 19.5
Missing	60.9%	0.9%	0.2%
Race
White	82.3%	85.0%	89.8%
African-American	13.0%	11.8%	7.4%
Other	4.7%	3.2%	2.8%
Cause of death
Anoxia	–	10.3%	3.9%
Cerebrovascular accident	–	32.3%	85.9%
Head trauma	–	53.7%	8.2%
CNS tumor	–	1.1%	0.6%
Other	–	2.7%	1.4%
First-year outcomes
Acute rejection
Antibody-treated	4.6%	4.5%	4.2%
Non-antibody-treated	8.8%	9.8%	11.6%

Percentages reflect proportions of recipients of a given donor type with the indicated trait (‘column percentages’). Means are provided as ± standard deviations.

Healthcare payments

There were strong associations between eGFR measured at 12 months post-transplant and subsequent healthcare expenditures. Multivariate regression analyses describing associations of baseline factors with payments in the second year after transplant for each donor type are presented in Table 2. The spline terms estimating eGFR effects are difficult to interpret numerically but are easily understood when displayed visually (Figure 1). Lower renal function, measured as eGFR, was a significant predictor of increased costs of care during the second year post-transplant for each donor type. Figure 1a displays the cost relationship as it varies with eGFR for recipients of SCD organs. Above 45 mL/min/1.73 m² the lower 95% confidence limit (LCL) bounds zero dollars, indicating that there is no significant difference in the marginal payments compared to the reference eGFR of 75 ml/min/1.73 m². As eGFR drops below 45 mL/min/1.73 m², eGFR is associated with significantly rising costs in an accelerating pattern. Compared to a reference eGFR of 75 mL/min/1.73 m², 1-year eGFR of 20 ml/min/1.73 m² in SCD recipients was associated with ∼ $17,500 higher adjusted payments in the second post-transplant year. An eGFR of 10 ml/min/1.73 m² in SCD recipients was associated with $27,437 (95% CI = 22,981–31,894) higher adjusted second year payments compared to eGFR of 75 ml/min/1.73 m². Our ability to identify the relationship between eGFR and marginal payments diminishes below ∼ 10 mL/min/1.73 m² due to a small sample of subjects who have a functioning graft at 12-months post-transplant at an eGFR less than 10 mL/min/1.73 m², as well as edge effects of the spline fitting methods.

Figure 1.  Marginal healthcare costs in the second and third years after transplant according to 12-month eGFR by cubic spline regression. Marginal payments are referenced to eGFR 75 ml/min/1.73 m². Item A includes 95% confidence limits for marginal payments across renal function levels among SCD recipients. Confidence limits are excluded from plots stratified by donor type for clarity. Basis knots for the splines were chosen to follow the K/DOQI ‘Stages of Chronic Kidney Disease’ starting at 15 mL/min/1.73 m² with additional knots at intervals of 15–150 mL/min/1.73 m². An additional knot at 22.5 mL/min/1.73 m² was found to significantly increase fit and was included in all models. Equations for the spline functions in SAS are provided in the Appendix. Significant second-year cost differences compared to the reference eGFR of 75 mL/min/1.73 m² emerged as 12-month eGFR declined below 40, 45, and 20 ml/min/1.73 m² among LD, SCD, and ECD recipients, respectively. Significant third-year cost differences compared to the reference eGFR of 75 mL/min/1.73 m² emerged as 12-month eGFR declined below 41, 48, and 20 ml/min/1.73 m² among LD, SCD, and ECD recipients, respectively.

Table 2.  Multivariable regression models for prediction of costs in the second year after transplant according to 1-year eGFR and other baseline factors.

	LD	SCD	ECD
Intercept	$53,465†	$28,930*	$12,266
eGFR at 1-year post-transplant	See Figure 1b	See Figures 1a and b	See Figure 1b
Joint Significance of eGFR^a	§	§	‡
Candidate/recipient variables
Recipient age (years/10)	$711.62†	$897.85§	$1572.90*
Male	−$634.11	−$964.12*	$545.90
Pre-transplant dialysis
Pre-emptive transplant	−$329.95	−$2419.80	$20,256
Natural log of (years of dialysis + 1)	$830.72	$1751.94	$16,951*
Body mass index (kg/m^2)	$296.97†	$84.33	−$308.39
> 20	−$208.84‡	−$85.78*	$193.32
Missing	$2785.01	−$390.57	−$1782.13
Cause of ESRD
Glomerulonephritis and diabetes	Reference	Reference	Reference
Hypertension	$1489.26	$1795.19†	−$301.18
Polycystic kidney disease	$53.50	−$108.97	−$8153.24*
Other	$636.16	$1729.31‡	−$2570.92
Diabetes	$21,279‡	$14,583§	$7470.74
Diabetes × Age (years/10)	−$2074.82§	−$825.25*	$41.06
Previous transplant	$1917.20*	−$536.67	$1243.13
Peak panel reactive antibody level
< 10	Reference	Reference	Reference
10–79	$1188.52	$1566.15‡	$1099.43
≥ 80	$3988.33*	$2626.77‡	$2402.26
Missing	−$331.05	$1909.83	$15,424†
HLA mismatches
Zero A, B, and DR	−$2220.17*	$398.68	$1180.36
Zero DR	−$401.64	−$1228.43†	−$3258.89
Serum albumin at listing (g/dl)	−$176.62	−$166.77	$2150.16
> 3.5	$61.08	−$492.31	−$71.53
Missing	−$696.14	−$3043.73	$8637.34
Race
White	Reference	Reference	Reference
African-American	$2068.07	$3117.04§	$3507.35
Other	−$2134.06	−$2042.31†	−$622.78
Year of transplant (1998)	$283.38	$297.24‡	$206.35
Donor variables
Age (in years)	$78.96	−$47.35	$226.71
> 18	−$50.28	$12.86	–
Missing	−$11,973	−$1248.48	–
Hypertension	–	$251.02	−$358.61
Cytomegalovirus positive	−$1326.98	$603.57	$1884.17
Natural log of weight (kg)	−$722.94	−$287.01	−$292.47
Weight missing	−$1866.78	−$4428.68	−$3424.20
Race
White	Reference	Reference	Reference
African-American	$238.96	$594.12	$2884.77
Other	−$2140.99	$1288.40	$1813.24
Donation after cardiac death	–	–	–
Cause of death
Anoxia	–	−$774.94	−$5026.16
Cerebrovascular accident	–	−$554.53	−$5100.33
Head trauma	–	−$864.51	−$5524.49
CNS tumor	–	$2456.28	−$3023.54
Other	–	Reference	Reference
First-year outcomes
Acute rejection
Antibody-treated	$3756.89†	$2960.60‡	$6590.22
Non-antibody-treated	$3626.64‡	$4101.57§	$1799.92

*p = 0.02–0.04; † p = 0.002–0.01; ‡ p = 0.0001–0.001; § p < 0.0001.

^aThe joint significance of eGFR was assessed using an F-test of the null hypothesis that the spline parameters (defined in Appendix 1) do not contribute significantly to the prediction of costs.

Figure 1b displays the relationship of 1-year eGFR between 10–80 mL/min/1.73 m² with second-year payments among LD, SCD, and ECD recipients compared to reference eGFR of 75 ml/min/1.73 m². A consistent pattern of accelerating costs at low eGFR but no significant effects at higher eGFR was observed across donor types, with particular similarity among LD and SCD recipients. Significant second-year cost differences compared to the reference eGFR of 75 mL/min/1.73 m² emerged as 12-month eGFR declined below 40, 45, and 20 ml/min/1.73 m² among LD, SCD, and ECD recipients, respectively.

Multivariate regression analyses describing the association of healthcare expenditures in the third year post-transplant with model factors for each donor type are presented in Table 3. Lower renal function, measured as eGFR at 12 months, was a significant predictor of increased healthcare payments during the third year post-transplant among LD, SCD, and ECD recipients (p < 0.0001). Although the ECD estimates were considerably less precise due to the relatively small sample, there were consistent patterns of accelerating third-year costs at low eGFR and among LD, SCD, and ECD recipients (Figure 1c). Significant third-year cost differences compared to the reference eGFR of 75 mL/min/1.73 m² emerged as 12-month eGFR declined below 41, 48, and 20 ml/min/1.73 m² among LD, SCD, and ECD recipients, respectively.

Table 3.  Multivariable regression models for prediction of costs in the third year after transplant according to 1-year eGFR and other baseline factors.

	LD	SCD	ECD
Intercept	$64,783†	$31,070†	$14,100
eGFR at 1 year post-transplant	See Figure 1c	See Figure 1c	See Figure 1c
Joint significance of eGFR^a	§	§	§
Candidate/recipient variables
Recipient age (years/10)	$469.71	$895.65§	$1148.77
Male	−$767.65	−$1154.84†	$1122.03
Pre-transplant dialysis
Pre-emptive transplant	−$14,679*	−$310.79	$20,637
Natural log of (years of dialysis + 1)	−$7687.23*	$1022.89	$16,300*
Body mass index (kg/m^2)	$291.77*	$120.90	$338.21
> 20	−$175.45†	−$101.75*	−$20.00
Missing	$3951.50	$878.16	$10,021
Cause of ESRD
Glomerulonephritis and diabetes	Reference	Reference	Reference
Hypertension	$1165.54	$1678.81†	$6046.92†
Polycystic kidney disease	−$1433.67	−$2096.66*	−$1826.02
Other	$614.64	$1837.72†	$2347.49
Diabetes	$14,807§	$12,795§	$21,672*
Diabetes × Age (years/10)	−$1172.24*	−$633.01	−$1814.62
Previous transplant	$153.36	$820.09	$4552.71
Peak panel reactive antibody
level
< 10	Reference	Reference	Reference
10–79	$1297.83	$1314.39†	$2908.60
≥ 80	$237.39	$763.98	$5481.69
Missing	−$1151.85	−$1.864.40	$10,860*
HLA mismatches
Zero A, B, and DR	−$1845.75	$500.85	−$5310.56
Zero DR	−$865.04	−$682.74	−$3.80
Serum albumin at listing (g/dl)	$723.73	$912.58	−$9739.79
> 3.5	−$433.03	$509.17	$4181.60
Missing	$169.79	$2305.29	$27,370
Race
White	Reference	Reference	Reference
African-American	$4886.45*	$4489.19§	$993.30
Other	−$461.83	−$2206.09*	−$781.80
Year of transplant	$439.53	$133.58	$390.85
Donor variables
Age (in years)	$194.22	−$37.59	$159.77
> 18	−$168.28	$22.09	–
Missing	$20.76	−$1577.57	–
Hypertension	–	$462.83	$1525.12
Cytomegalovirus positive	−$765.18	$910.65*	$2741.27
Natural log of weight (kg)	$2615.46	−$1041.26	−$1909.64
Weight missing	$12,586	−$8317.37*	$12,591
Race
White	Reference	Reference	Reference
African-American	−$1277.03	$1578.30*	$4851.40
Other	−$2611.81	$3732.13†	−$1437.43
Cause of death
Anoxia	–	$495.99	−$10,365
Cerebrovascular accident	–	−$333.80	−$8969.01
Head trauma	–	−$498.53	−$10,232
CNS tumor	–	$2243.79	−$22,046
Other	–	Reference	Reference
First-year outcomes
Acute rejection
Antibody-treated	$2167.00	$3815.98§	$8104.72*
Non-antibody-treated	$3490.16†	$2883.02§	$1132.41

*p = 0.02–0.04; †p = 0.002–0.01; ‡p = 0.0001–0.001; §p < 0.0001.

^aThe joint significance of eGFR was assessed using an F-test of the null hypothesis that the spline parameters (defined in Appendix 1) do not contribute significantly to the prediction of costs.

Many covariate factors were significantly associated with Medicare payments in the second and/or third year post-transplant. These were most precisely estimated in the largest sub-group, SCD recipients in the second year post-transplant, among whom costs increased with older recipient age, female compared to male gender, higher sensitization levels, and the presence of DR mismatches (compared to 0 DR mismatches). The only factor with a significant cost effect comparable to impaired 12-month eGFR less than 30 mL/min/1.73 m² was pre-existing diabetes. Factors with highly significant cost effects comparable to a 12-month eGFR between 30–45 mL/min/1.73 m² were recipient age, certain causes of ESRD, elevated peak panel reactive antibody, African American recipient race, and acute rejection.

Discussion

Better understanding of the relationships of post-transplant renal function with healthcare costs among kidney transplant recipients is vital to the economic assessment of potential treatment strategies that may impact renal allograft function. We examined national registry data for a large cohort of US transplant recipients to quantify associations of eGFR at the first transplant anniversary with actual Medicare payments for healthcare services in the second and third years after transplant. Flexible functions known as cubic splines16 were employed to allow the shape of the cost relationships to vary non-linearly across levels of eGFR. We found that eGFR measured at 1-year after kidney transplant is significantly associated with subsequent Medicare payments, but this relationship is not linear. Marginal costs begin to increase as eGFR falls below 45 mL/min/1.73 m² and rise quickly with further decline in renal function. Compared to the reference eGFR of 75 mL/min/1.73 m², 1-year eGFR of 10 ml/min in SCD recipients was associated with increases of ∼ $28,000 in payments estimated from our multivariate analyses in both the second and third post-transplant years. Patterns were particularly similar among recipients of LD and SCD transplants, while estimates among ECD recipients were less precise due to the smaller samples. However, the significance of eGFR as a cost predictor among all donor types demonstrates the importance of renal function following kidney transplantation for post-transplant economic outcomes, regardless of the source of the donated organ.

Quantification of the costs associated with health states offer valuable insights for economic evaluations including cost-benefit analyses of standard and emerging therapies. An important observation from these data is that the marginal cost implications of renal insufficiency among SCD and LD transplant recipients surviving to the first anniversary begin as eGFR declines below ∼ 40–45 mL/min/1.73 m², an observation consistent with a recent study that considered eGFR in terms of functional stages rather than small units of change6. Similar to these findings in transplant recipients, covariate-adjusted healthcare charges in patients with native CKD from autosomal dominant polycystic kidney disease were stable in CKD stages 1–3, but then began to rise precipitously as eGFR declined less than 30 ml/min/1.73 m² 3. Smith et al.2 found that, while care of native CKD patients was more expensive than that of controls without CKD in an HMO setting, provider-side costs were stable across pre-end stage CKD stages 2–4.

The increase in healthcare charges for kidney transplant recipients with eGFR < 45 ml/min/1.73 m² is important. Advancing CKD of both native and transplanted kidneys increases the burdens of related complications such as anemia, hypertension, bone and mineral metabolism disorders, and cardiovascular disease18–21. Medicare payments increased sharply among transplant recipients with eGFR < 30 ml/min/1.73 m² in the current study. While the study sample was selected to have a functioning allograft at the first transplant anniversary, stage 4 CKD (eGFR 15–29 ml/min/1.73 m²) is the functional level at which many transplant recipients begin preparing for dialysis or re-transplantation. Healthcare costs have been shown to rise with proximity to renal replacement therapy in native CKD patients. For example, St. Peter et al.22 reported increases in Medicare/Medicaid costs from $993 per-patient per-month 2 years before dialysis to $6300 per-patient per-month 6 months after dialysis initiation in an incident dialysis population (1997 dollars). Prior analyses have considered the impact of modifying the rate of decline to ESRD among CKD patients. Trivedi et al.23 estimated that reducing the rate of GFR decline by 10% and 30% in every American with GFR ≤ 60 mL/min/1.73 m² would yield gross direct cumulative healthcare savings over a decade of $18.56 and $60.61 billion, respectively. This reflects both the incremental cost of caring for CKD and the potential to avoid ESRD and dialysis.

For kidney transplant patients, renal function preservation has important economic consequences for medical care and possible re-transplantation. The costs of complete kidney allograft failure have been quantified as ∼ $100,000 in excess Medicare payments from the immediate period of graft failure through the following year compared to maintaining a functioning graft5. Progression to ESRD also commonly results in re-listing of the patient for a second transplant. Repeat transplant has been shown to be both more expensive and less likely to be successful than primary transplants, particularly when performed with allografts from ECD organs24. By diminishing the number of patients who require re-transplantation, preservation of renal function will allow the limited number of available allografts to be used most cost-effectively in first-time recipients. The current study advances understanding of the cost implications of pre-end stage renal dysfunction after kidney transplant across a spectrum of function levels, quantifying accrual of economic impact across advancing degrees of renal insufficiency. However, the estimates provided here may be conservative in that they do not include the opportunity costs associated with use of future organs in the context of re-transplantation.

To date, accepted end-points in clinical immunosuppression trials among kidney transplant recipients have incorporated graft failure, patient death, and, more recently, acute rejection25. Emerging data support post-transplant renal function as a strong predictor of ‘hard end-points’ of patient and graft survival, suggesting that eGFR may be a useful trial end-point6,26,27. Further, cost-effectiveness is increasingly considered along with clinical effectiveness as insurers and providers consider adoption of alternative treatment strategies. The current data suggest that post-transplant eGFR may be a useful metric for discriminating the economic impact of care strategies that differentially affect renal function.

Limitations of this study include the retrospective design. Although this is a contemporary sample of renal transplant recipients, it is possible that changes in post-transplant management may alter long-term cost implications of renal function. We employed eGFR as a surrogate for measured GFR, which has likely introduced statistical noise in the estimates. The typical consequence of statistical noise is a reduction in the significance of the estimated effects. Therefore, it is possible that the relationship between measured GFR and post-transplant healthcare costs could be estimated with more statistical precision, and that significant differences in costs could be detected at thresholds closer to the reference eGFR level than identified here. Although the use of measured GFR might provide more accurate estimates of the relationship of renal function with costs, measured GFR is not commonly obtained in the routine care of kidney transplant patients and, thus, is not available for a large historical cohort study such as presented here. The study sample was limited to Medicare beneficiaries and findings may differ under other reimbursement systems such as private insurance. However, as Medicare the dominant payer for patients of all ages with end-stage renal disease in the US including transplant recipients, these findings are likely to be highly generalizable. The study sample was also limited to survivors with allograft function at the first transplant anniversary, and eGFR–cost relationships may differ in the early and very late periods after transplant. With respect to our regression approach, alternatives to our OLS models, such as regressions estimating the determinants of the natural log of Medicare payments, may be more efficient but also may produce biased estimates and are difficult to interpret. Because we have access to cost data for very large samples, we employ the unbiased estimator. Our past work has demonstrated nearly identical results with OLS cost regression and regressions on the natural log of Medicare payments28, and OLS has become our standard in analyses of the economic impact of complications in transplantation6,12,29.

Conclusions

In conclusion, we found that eGFR at 1-year post-transplant is significantly associated with Medicare expenditures during the second and third year after kidney transplantation. Significant patterns hold when separately estimated across donor types. This evidence is relevant to economic evaluations of new interventions that could prevent or delay kidney dysfunction after transplantation. Strategies that slow the decline in transplant renal function have the potential to generate substantial reductions in healthcare costs.

Transparency

Declaration of funding

Support for data research staff time was provided by a grant from Bristol-Myers Squibb. The manuscript does not include discussion of any pharmaceutical product, other healthcare product, or off-label use of medications. The analyses, interpretation, medical writing, and reporting of these data are the responsibility of the authors. The roles of the authors in this work are as follows: M.A.S. participated in study design, data analysis, and writing of the paper. A.G., D.A., G.L., and K.L.L. participated in study design, interpretation, and writing of the paper. All authors agreed to publish the paper.

Declaration of financial/other relationships

M.A.S., A.G., K.L., and D.A. have disclosed that they received support from a grant from Bristol-Myers Squibb. G.L. is an employee of Bristol-Myers Squibb.

Acknowledgments

The data reported here have been supplied by the US Renal Data System (USRDS). The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the US Government.