The Economics of Renoprotective Therapy in Advanced Diabetic Kidney Disease
A budget-impact model illustrates the economic outcomes associated with renoprotective therapy on privately insured patients with advanced kidney disease.
Approximately one-third of patients in the United States with chronic kidney disease (CKD) have type 2 diabetes mellitus.1 Data indicate that up to 58% of these patients fail to receive recommended renin-angiotensin-aldosterone system (RAAS) inhibition therapy despite proven benefits in slowing CKD progression and delaying progression to end-stage renal disease (ESRD).1,2 The cost to private insurance plans of the failure or inability to place these patients on appropriately dosed renoprotective therapy has not been established.
Our study sought to estimate the potential benefits of one form of RAAS inhibition therapy using angiotensin II receptor blockers (ARBs) by modeling a well-studied clinical trial cohort. We created a hypothetical patient cohort similar to the treatment and placebo arms of one of the landmark clinical trials establishing the benefit of ARB therapy in diabetics, the Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan (RENAAL) trial.3 Patients with macroalbuminuria similar to those studied in the RENAAL trial comprise approximately 6% of patients with stage 3 CKD and 42% of patients with stage 4 CKD.4 We modeled CKD progression, ESRD progression, mortality, and total healthcare costs over a 10-year period under 2 disease progression risk scenarios with private insurance payment.
A budget-impact model (BIM) was developed to assess the economic burden of diabetic kidney disease on US health insurance plans and the potential direct savings (medical and drug) to these plans from the use of ARB therapy. Data sources included the following:
Two double-blind, placebo-controlled clinical trials initially established the value of ARB therapy on hard renal outcomes in persons with diabetes and advanced kidney disease: RENAAL and Irbesartan Diabetic Nephropathy Trial (IDNT), both published in 2001.5,6 Both trials examined a composite primary end point, including progression to ESRD, in similar populations. RENAAL’s definition of ESRD progression (initiation of long-term dialysis or renal transplantation) corresponded to a discrete medical resource end point, while IDNT’s definition also included a serum creatinine end point. Accordingly, we selected RENAAL as the representative for efficacy of ARB therapy in this study and the basis of the BIM.
To model the risk and timing of disease progression, mortality, healthcare costs, and hyperkalemia-related inputs, we conducted a search of the peer reviewed literature and clinical guidelines in Medline, Cochrane, ISPOR, and National Guideline Clearinghouse databases, with preference to sources presenting results for patients with diabetes and/or macroalbuminuria, not yet of Medicare age. Primary data sources and uses are summarized in
; search and selection details are provided in the
(available at www.ajmc.com).
Progression of CKD, Including Mortality
The most widely accepted analysis of progressive renal disease is the set of meta-analyses covering more than 1.5 million patients presented by Levey and colleagues at the KDIGO Controversies Conference in 2009.7 This work confirmed that the critical factors determining renal risk are disease stage, as measured by estimated glomerular filtration rate, and albuminuria; it furnished the relative and absolute risk probabilities we used to predict the progression between stages 3a, 3b, and 4, and into ESRD, as well as pre-ESRD mortality risk. Annual mortality risk during ESRD treatment was computed from United States Renal Data System (USRDS) cohort survival statistics.1 We defined the base case for the BIM as a hypothetical cohort of patients of similar size and disease characteristics as the RENAAL clinical study population, making adjustments to our foundational data sources to account for the high levels of proteinuria in the RENAAL population (median urinary albumin:creatinine ratio [UACR] >1200)3 (see eAppendix for details). The conservative case scenario represents a similarly sized cohort but makes no adjustment for the elevated proteinuria in this patient population, instead using the absolute and relative risk of disease progression from all persons with UACR equal to or greater than 300 in the KDIGO model.7 Thus, the conservative case represents a less acutely ill CKD population than the base case.
Efficacy of Therapy
To model outcomes in treated patients, we reduced the absolute risk of ESRD progression by 29% and the absolute risk of CKD progression by 25%, as observed in RENAAL patients on losartan.8 We modeled no direct impact on mortality from losartan therapy.5 However, differences in mortality emerge in the BIM due to slower progression of losartan-treated patients into higher-mortality CKD stages and ESRD.
A private insurance plan’s annual cost for a member with CKD roughly doubles with each progressive stage of CKD,9,10 is higher for patients with diabetes11; among patients with diabetes, is higher for those with macroalbuminuria12; and spikes in the months just before and after progression to ESRD.10,11 Annual estimated baseline costs by CKD stage, used for both the base and conservative cases, are shown in
(stage 3: $15,000 in 2001 dollars10 *1.27 [diabetic:total]11 *1.19 [macroalbuminuria:total among diabetics],12 inflated13 = $34,420; allocated $31,597 stage 3a [79%], and $44,867 stage 3b [21%],7 to equalize the percentage cost increase. Stage 4: $28,000,10 similarly adjusted = $64,251. ESRD: cost of stage 4 + annualized cost of progression to ESRD in a patient with diabetes and macroalbuminuria [$56,745 in 2009 dollars,12 inflated to $63,009] = $127,350) (details in the eAppendix). To the baseline, we added an additional $52,915 in the year preceding ESRD, derived from Joyce’s findings of cost acceleration at this point.11
Hyperkalemia (HK), defined as a serum potassium level exceeding the upper limit of normal for a given laboratory (generally >5.0 mmol/L) is a known side effect of RAAS inhibition therapy, including ARB therapy. RENAAL investigators reported that 22.8% of losartan-treated patients experienced potassium (K) greater than 5.5 mmol/L versus 5.1% of placebo patients.5 Einhorn reported 30.8% annual incidence of lab values of K greater than 5.5 mmol/L among 35,744 community-setting diabetics with stage 3 to stage 5 CKD,15 which we adjusted for the frequency with which HK has been noted among similar patients as a primary16 or secondary17 reason for medical care, estimating that 28% of patients with excessive potassium lab readings received explicitly related healthcare services (3.18% treatment event incidence with HK primary diagnosis16/30.8% lab value incidence15 * 39% of HK-related events in which HK diagnosis was primary17 + 12.12% HK secondary event rate16,17/30.8% * 61% of HK-related events in which HK diagnosis was secondary17 = 28%).
For patients in the treatment arm, we calculated an annual cost for generic losartan therapy of $168, based on an Internet-accessed retail price of $14 per month for losartan 100 mg. We ceased applying cost for losartan therapy when patients progressed to ESRD, as any continuation after that point is for purposes other than preventing kidney failure. We added the cost of generic losartan therapy and the costs of treating hyperkalemia to the baseline costs in the BIM.
Using the inputs described above, we modeled distribution by CKD stage of the RENAAL losartan and untreated study arms at trial baseline, and projected disease progression and mortality annually for 10 years, assuming ongoing effectiveness of therapy at levels demonstrated in RENAAL. We applied cost in 2012 dollars to the cohort of patients at each stage in each study arm, with adjustments for short-term costs surrounding progression to ESRD, the differential cost of HK, and the cost of therapy. Average annual cost per patient in each year was calculated as the sum of costs for treating patients at each stage (baseline plus relevant additions) divided by the number of surviving patients; all costs were normalized to 2012 dollars.13 All modeling and nonstatistical analyses were conducted in Microsoft Excel (2007). Differences between cumulative proportions in the 2 study arms of the modeled population were evaluated using a χ2 test, and carried out using SAS/STAT software (version 9.2).
For the base-case scenario, the average annual total healthcare cost to a private insurance plan for a cohort of patients matching the RENAAL population was estimated to be $58,979 per patient treated with losartan and $62,386 per patient without ARB therapy (ie, untreated) in year 1. Average savings in healthcare costs during the first year (net of the cost of losartan therapy and the cost of treating increased hyperkalemia) was estimated at $3408. By year 3, average annual savings reach $6323 per patient with treatment. After 10 years, annual costs equal $83,265 per person on losartan versus $94,255 per untreated patient, for annual savings with treatment of $10,990 per patient. Under the conservative case (ie, less acutely ill patients), the average savings in healthcare costs (or the net of cost of therapy) during the first year are estimated at $1154 per patient. By year 3, average annual cost is $57,180 for patients taking losartan versus $59,864 for untreated patients, a savings of $2685. By year 10, the losartan cohort costs, on average, $64,776 per patient compared with $70,435 per untreated patient, for an annual savings of $5659 per patient (
Under the base case, assuming a patient cohort of diabetic CKD patients similar in number to the RENAAL trial (about 750 treated patients), cumulative cost savings with
losartan treatment reach $11.1 million by year 3 and $18.4 million by year 5, compared to $5.3 million in year 3 and $9.5 million in year 5 for the conservative case. Expected cumulative savings by year 10 reach $24.4 million in the base case and $16.4 million under the conservative case.
In the first year, 46 losartan patients and 65 untreated patients were expected to progress to ESRD under the base case. The cumulative number of patients presenting with ESRD within 3 years rose to 129 patients (17.2%) with losartan versus 182 patients (23.9%) without losartan therapy (odds ratio [OR], 0.66; 95% CI, 0.51-0.85, as shown in Figure 3). Outcomes were expected to continue to diverge in subsequent years, with 76 fewer losartan-treated patients in the base case progressing to ESRD after 5 years: 201 losartan patients (26.8%) versus 277 untreated patients (36.4%) (OR, 0.64; 95% CI, 0.51-0.80). Over 10 years, ESRD progression is expected in 333 patients on losartan (44.4%) versus 434 untreated patients (56.9%) (OR, 0.60; 95% CI, 0.49-0.74). In the conservative case, ESRD progression after 3 years is expected in 62 patients on losartan (8.3%) and 89 (11.6%) untreated patients (OR, 0.68; 95% CI, 0.48-0.96). By year 10, disease progression is expected in 174 patients on losartan (23.1%) and 239 (31.4%) untreated patients (OR, 0.66; 95% CI, 0.53-0.83).
In the base case, expected deaths in this cohort exceed 50% over 10 years, with or without treatment. Within 3 years, 101 losartan patients (13.5%) and 113 untreated patients (14.8%) are expected to expire (OR, 0.89; 95% CI, 0.67-1.19). Cumulative mortality at 5 years is 186 losartan patients (24.8%) versus 213 untreated patients (27.9%) (OR, 0.85; 95% CI, 0.67-1.07). After 10 years, deaths reach 51.8% (389 patients) with losartan versus 58.4% without ARB therapy (445), and the difference achieves significance (OR, 0.77, CI, 0.62-0.94). In the conservative case, mortality differences within 10 years for 310 losartan patients (41.2%) versus 345 untreated patients (45.2%) are not significant (OR, 0.85; 95% CI, 0.69-1.04).
CKD progression is mitigated for many patients by the appropriate use of RAAS inhibitors, yet the proportion of CKD patients receiving this renoprotective therapy remains relatively low despite clinical guidance.1 Our study sought to estimate the potential benefits of one form of RAAS inhibition therapy—angiotensin II receptor blockers (exemplified by losartan)—illuminating the economic impact of CKD both before and after the progression to ESRD.
Economic analyses of CKD often focus on the financial benefit of avoiding ESRD, for which the increase in costs is large and easily substantiated. Our study suggests that this common approach may understate the full cost burden of CKD progression at the earlier stages likely to be reached before patients age to Medicare coverage. Indeed, systematic searching for data distinguishing costs for CKD patients in stages 3 and 4 with private insurance, without excluding end-of-life costs, yielded cost estimates that are higher than more commonly cited sources requiring continuous insurance coverage,1 or evaluating only ESRD avoidance as an economic benefit.18
Our analysis shows that even under conservative assumptions of disease progression, the impact of CKD among the privately insured is significant, both clinically and economically, particularly among patients who do not receive appropriately dosed RAAS therapy. In our base case cohort model, net savings of treatment costs averaged more than $3400 for every patient on losartan therapy in the first year, and increased each year, reaching annual savings of $10,990 per surviving patient in year 10. Importantly, expected cumulative savings peak at $24.7 million in year 9 for this cohort of patients similar in size and severity to the RENAAL patient population (about 750 treated patients). By year 10 in the base case, the cost of treating the larger number of survivors in the losartan study arm finally outweighs the per patient savings with treatment, and cumulative savings decline to $24.4 million, as shown in
. In our more conservative scenario, savings average $1154 for every patient on losartan therapy in the first year, reaching savings of $5,659 per surviving patient in year 10 and cumulative savings exceeding $16 million over 10 years. In either case, the potential cost savings with therapy are substantial and associated with reduced mortality and increased quality of life, as patients avoid or delay dialysis. With the increasingly obese US population developing type 2 diabetes mellitus at younger ages, these findings raise important considerations for private insurers in the United States.
RAAS inhibition therapy is not expensive. We estimated $168 per patient per year for generic losartan. Although the reasons for nontreatment and undertreatment with RAAS inhibitors are multifactorial,1,19-22 one well-recognized limiting factor is concern about HK.8 Potassium levels in both the RENAAL and IDNT trials were significantly higher among patients treated with RAAS inhibitors, and elevation in potassium level has been noted in much of the literature on RAAS inhibition therapy.15,23 Preventing or treating HK may, therefore, be one of the keys to fulfilling the potential of RAAS inhibitors. Emerging therapies designed to manage potassium levels show promise in this regard.24 Trials evaluating these pharmacotherapies should address the potential for optimization of RAAS inhibitors, as well as potassium control itself.
Results are specific to persons with advanced diabetic kidney disease (ACR >300) without heart failure treated with and without losartan, and cannot be generalized to broader populations. Fewer African Americans were represented in Nichols’ study (3.5%) and in the CKD meta-analysis contributing key results to the Levey model (7.5%)14 than in RENAAL (13.8%) or among ESRD patients in the United States (27.8%). Because CKD is a widely studied clinical condition, we focused on the use of methodologically sound and frequently cited studies using large, well-known data sources of relevance to private insurers rather than conducting an original systematic review or a meta analysis of the literature. The duration and statistical power of the RENAAL study was insufficient to draw specific conclusions about the effect of therapy on mortality; thus, our model’s calculated mortality benefit was based exclusively on estimated deaths tied to CKD progression and ESRD.
Our linear assumption of losartan efficacy necessarily extends beyond the time studied in the clinical trial. Available data on the relationship between hyperkalemic lab values and actual treatment cost was particularly scant, warranting further study. Also, reliance on the RENAAL study as representative of ARB therapy necessarily excludes the extensive accumulated research on the full family of RAAS inhibition therapies; the average cost per patient of ARB therapy may be higher, particularly if nongeneric drugs are utilized. Estimated savings to an individual private insurer could be lower than predicted over time as patients move into Medicare coverage or switch to another commercial plan. Conversely, cost savings may be higher in patients using RAAS inhibitors if serum potassium can be managed; Miao et al’s post hoc adjustment of RENAAL results for serum potassium suggested that losartan’s renoprotective effect could improve from 21% to 35%.25 Thus, our study should be considered a first step in eliciting further investigation of diabetic CKD in privately insured populations.
ACE inhibitor and ARB therapies have been shown to delay progression of advanced diabetic kidney disease. Our BIM suggests that, at private insurance payment rates, the use of ARB therapy reduces average annual healthcare costs for patients with advanced diabetic kidney disease between $1150 and $3400 in the first year. These savings rise to between $5600 and $11,000 per surviving patient in year 10, provided that HK or other factors do not prevent patients from remaining on therapy. We hope our findings will stimulate greater sharing of non-Medicare cost data on patients with diabetes and renal disease, and direct investigation into the benefits/risks associated with renoprotective therapies.