Sevelamer Is Cost-Saving vs. Calcium Carbonate in Non-Dialysis-Dependent CKD Patients in Italy: A Patient-Level Cost-Effectiveness Analysis of the INDEPENDENT Study

Hyperphosphatemia, above-normal (i.e. >4.3 mg/dl) levels of phosphorus in the blood, is a common consequence of chronic kidney disease (CKD) that has been associated with an increased risk of cardiovascular morbidity, cardiovascular mortality, all-cause mortality, and faster progression to the need for dialysis in patients with non-dialysis-dependent CKD (NDD-CKD) [1,2,3,4,5,6,7]. The current National Kidney Foundation Kidney Disease Improving Global Outcomes (KDIGO) clinical practice guidelines recommend dietary restriction of phosphate intake and the use of phosphate binders to maintain phosphorus within normal range for patients with CKD [1]. The use of phosphate binders has been associated with statistically significant improvements in cardiovascular and all-cause survival in patients at all stages of CKD [8,9,10,11] and delays in the progression of NDD-CKD patients to dialysis [7,11].

In most jurisdictions, including Italy, current options for phosphate binders include calcium-based binders (CBBs) such as calcium carbonate, calcium acetate, and combination calcium acetate plus magnesium carbonate and non-CBBs such as sevelamer hydrochloride or sevelamer carbonate (sevelamer) and lanthanum carbonate (lanthanum). Each of these phosphate binders has been shown in clinical trials, systematic reviews, and meta-analyses to be efficacious in lowering serum phosphorus in patients with CKD [12,13,14,15,16,17]. Several studies show that some of the calcium from CBBs binders is absorbed into the bloodstream and possibly deposited in small blood vessels causing vascular calcification and organ damage [18,19,20] and leading to increased risk of mortality. CBBs are often recommended and prescribed for first-line treatment of hyperphosphatemia because they are the least costly of the available phosphate binder options. The KDIGO guidelines, for example, caution against the prescription of phosphate binders that are ‘less cost-effective' than CBBs in the absence hypercalcemia [1].

Recently, the INDEPENDENT-CKD study, a multicenter, randomized, open-label study, evaluated the impact of sevelamer versus calcium carbonate on all-cause mortality (primary outcome), the inception of dialysis (secondary outcome), and the composite outcome of all-cause mortality and inception of dialysis (secondary outcome) in patients with stage 3-4 CKD in Italy [11]. Over the duration of the study (3 years), patients randomized to sevelamer (n = 107) had significantly lower all-cause mortality (HR adjusted for baseline covariates: 0.35, 95% CI 0.15-0.76; HR adjusted for baseline and time covariates: 0.36, 95% CI 0.15-0.83; p < 0.01) and a longer duration to dialysis inception (HR adjusted for baseline covariates: 0.49, 95% CI 0.29-0.81; HR adjusted for baseline and time covariates: 0.77, 95% CI 0.45-1.34; p > 0.05) than patients randomized to calcium carbonate (n = 105).

This study utilizes patient-level data from the INDEPENDENT-CKD study to conduct a cost-effectiveness analysis (CEA) of sevelamer versus calcium carbonate in patients with NDD-CKD.

The CEA of sevelamer was performed using individual patient-level data from the INDEPENDENT-CKD study, a randomized, open-label clinical trial conducted in Italy. The study design, patient enrollment, and clinical results of the INDEPENDENT-CKD study are described in detail elsewhere [11]. Briefly, a total of 239 eligible patients were randomized in a 1:1 fashion to open-label treatment with either 1,600 mg/day sevelamer hydrochloride or 2,000 mg/day calcium carbonate (clinicians were free to increase the dose of phosphate binder thereafter). Patients were followed for 3 years or until all-cause death (the primary study end point). Inception of dialysis and the composite end point of all-cause mortality or inception of dialysis were also investigated as secondary study end points.

For the CEA, patient-level data from the INDEPENDENT-CKD study for time to death from all causes was used to calculate life years (LYs), and time to initiation of dialysis was used in the calculation of costs. Mean per-patient daily dose of sevelamer and calcium carbonate was also taken from the INDEPENDENT-CKD study and was included in the calculation of costs. Since the INDEPENDENT-CKD study did not collect data regarding quality of life or patient utility, LYs were used to calculate the incremental cost-effectiveness ratio (ICER). The CEA was performed from the perspective of the Italian NHS and included only direct medical costs. The time horizon for the analysis was 3 years, as treatment and follow-up data from the INDEPENDENT-CKD study were available for this period. No discounting was applied to either outcomes or costs. Bootstrapping was employed to estimate 95% confidence intervals around the point estimates for LYs, total costs, and the ICER.

Resource use and costs captured in the economic evaluation reflect direct medical costs associated with sevelamer and calcium carbonate usage, hospitalizations, and dialysis for those patients who initiated dialysis during the clinical trial. Other costs such as those associated with outpatient visits, concomitant medications, and adverse events were not included in the analysis. All costs were expressed in 2012 euros.

Clinicians in the INDEPENDENT-CKD study were free to adjust the dose of sevelamer and calcium carbonate in order to maintain serum phosphorus concentrations between 2.7 and 4.6 mg/dl for patients with stage 3-4 CKD, and between 3.5 and 5.5 mg/dl for patients with stage 5 CKD. Patient-level data regarding the daily dose of sevelamer and calcium carbonate at baseline and at 12 and 24 months of follow-up were used to inform sevelamer and calcium carbonate usage during the first, 2nd and 3rd years of the CEA, respectively. Assuming an 800-mg tablet for sevelamer and a 500-mg tablet for calcium carbonate, the daily number of tablets ranged from 2.0 to 4.0 and 4.0 to 8.0, respectively. The cost per tablet for sevelamer (EUR 0.75) and calcium carbonate (EUR 0.0165) was provided by the in-hospital pharmacy of Solofra Hospital, Avellino, Italy, and FARMADATI.

Data regarding hospitalizations were not collected in the INDEPENDENT-CKD study. However, because hospitalizations represent the largest component of resource utilization in patients with CKD [21] and because sevelamer has been shown previously to reduce the risk of hospitalizations in patients with CKD [22], it was deemed important to capture the frequency and cost of hospitalizations in the CEA. Post hoc to the INDEPENDENT-CKD study, a retrospective chart review of the patients included in the INDEPENDENT-CKD study was conducted to collect data regarding hospitalizations. Specific data for each hospitalization, including length of stay (LOS) and discharging ward, were collected for each patient from the point of randomization to the point of study completion (3 years) or death. Unit costs for hospitalizations were taken from the Italian diagnosis-related group tariffs and were applied, according to discharging ward, to each hospitalization for each patient to arrive at a total cost of hospitalization over the 3 years. The unit costs for a hospitalization discharged from the intensive care unit (ICU), nephrology ward, or cardiology ward was EUR 3,730, 3,746, and 3,694, respectively. The analyses assumed no differences in the unit cost of hospitalization between the treatment groups.

Patient-level data regarding the initiation of dialysis were available from the INDEPENDENT-CKD study, and the number of dialysis sessions occurring until either death or completion of 3 years of follow-up was collected retrospectively through patient's charts. The unit cost for a hemodialysis session (EUR 176.98) was taken from regional ambulatory tariff.

Categorical data were reported as frequencies and percentages, and continuous data were reported as mean ± standard deviation. Statistical significance between treatment groups was determined using t test and Fisher exact test for continuous and discrete variables, respectively; significance was set at 5% for all tests.

Total LYs and total costs over the 3-year period of the INDEPENDENT-CKD study were calculated for each patient from which mean LYs and mean costs were calculated and reported for sevelamer versus calcium carbonate. The cost-effectiveness of sevelamer relative to calcium carbonate was expressed as the incremental cost per LY gained and was calculated as the difference in mean cumulative 3-year costs between the two treatment groups divided by the difference in mean cumulative 3-year LYs.

Parametric bootstrapping methods, which allow for the estimation of various statistics such as confidence intervals using traditional parametric formulas [23], were used to estimate 95% confidence intervals around the point estimates for LYs, total costs, and the ICER. Briefly, the original dataset was used as a pool from which 10,000 new datasets of the same size were randomly drawn with replacement. Bootstrap estimates of the incremental LYs and incremental costs were calculated for each new dataset. From the bootstrap samples, the net monetary benefit statistic was calculated along with the probability that sevelamer was cost-effective compared with calcium carbonate over a range of willingness to pay (WTP) thresholds (e.g. maximum WTP for 1 LY gained). A cost-effectiveness acceptability curve (CEAC) was plotted using the estimates of the probability that sevelamer was cost-effective over the range of WTP thresholds.

Since the time to inception of dialysis between patients treated with sevelamer and patients treated with calcium carbonate was not statistically different in the INDEPENDENT-CKD study [11], we conducted a separate economic evaluation that excluded any impact of dialysis inception. The analysis included only those patients (sevelamer: n = 76; calcium carbonate: n = 63) who did not progress to dialysis within the 36-month time period of the clinical trial. The reasons for performing this subgroup analysis were twofold: (1) to evaluate the potential of an early phosphate binder treatment without the potential confounder of dialysis inception, and (2) to confirm the result in a more homogeneous, albeit smaller sample of patients.

Of the 239 patients enrolled, 27 were excluded from the analysis due to loss during follow-up [11]. The remaining 212 patients (sevelamer: n = 107; calcium carbonate: n = 105) were included in the primary efficacy analysis of the INDEPENDENT-CKD study and were included in this CEA.

Key demographic and disease background information for the patient population is summarized in table 1. A greater proportion of patients randomized to sevelamer demonstrated coronary artery calcification (62.6%) compared to patients randomized to calcium carbonate (47.6%; p = 0.02). There were no significant differences between treatment groups in the use of concomitant medications such as statins, calcium channel blockers, beta-blockers, ACE inhibitors/angiotensin II receptor antagonists, or native or active vitamin D (not shown).

Clinical Outcomes

Table 1 summarizes the trial-based LYs, hospitalizations, and dialysis associated with sevelamer and calcium carbonate for the base case analysis over 3 years. A total of 12 (11%) deaths occurred among sevelamer patients versus 22 (21%) deaths among calcium carbonate patients (p = 0.009); the rate of all-cause mortality was significantly less frequent among patients randomized to sevelamer (p < 0.01) [11]. The mean number of LYs per patient treated with sevelamer was 2.89 ± 0.35, while the mean number of LYs per patient treated with calcium carbonate was 2.83 ± 0.41; the gain of 0.06 (95% CI -0.04 to 0.16) LYs for sevelamer was not statistically significant at p = 0.23. A nonsignificant trend towards a lower rate of dialysis inception was observed among patients randomized to sevelamer (adjusted HR = 0.77; 95% CI 0.45-1.34). A total of 31 (29%) sevelamer patients initiated dialysis at some point during the study and underwent a total of 4,804 dialysis sessions, while a total of 42 (40%) calcium carbonate patients initiated dialysis and underwent a total of 6,957 dialysis sessions (p < 0.05 for both the percentage of patients initiating dialysis and the number of dialysis sessions).

The total number of patients admitted to hospital in the sevelamer group (n = 33; 31%) was significantly fewer than those admitted to hospital in the calcium carbonate group (n = 72; 69%; p < 0.001). The total number of hospitalizations was also significantly fewer for patients treated with the sevelamer than for patients treated with calcium carbonate (72 hospitalizations for sevelamer vs. 187 hospitalizations for calcium carbonate; p < 0.001). This trend was significant at the p < 0.001 level across all ward types: nephrology (59 for sevelamer vs. 141 for calcium carbonate), cardiology (10 vs. 33, respectively), and ICU (3 vs. 13, respectively).

Economic Outcomes

Table 2 summarizes the mean per-patient phosphate binder, hospitalization, dialysis, and total costs associated with sevelamer and calcium carbonate for the base case analysis over 3 years. The mean per-patient cost of sevelamer (EUR 2,402.46) was higher than that of calcium carbonate (EUR 99.70; p < 0.0001). The mean cost per patient for all hospitalizations was EUR 4,138 lower for patients treated with sevelamer than for those treated with calcium carbonate (p < 0.001). By ward type, mean costs per patient were lower for patients treated with sevelamer: EUR 2,965, 816, and 357 lower for hospitalizations to the nephrology, cardiology, and ICU wards, respectively. Any differences in the mean costs between treatment groups were derived solely from clinical data demonstrating a reduction in the mean number of hospitalizations with sevelamer. The cost savings associated with the reduction of hospitalizations for patients treated with sevelamer more than offset the higher acquisition cost for sevelamer.

Treatment with sevelamer was also associated with lower dialysis costs. For patients treated with sevelamer, the mean cost of dialysis per patient over 3 years was EUR 7,945 compared to EUR 11,726 for patients treated with calcium carbonate (p = 0.064). Taken together, the total mean cost per patient for patients treated with sevelamer (EUR 12,863 ± 15,790) was EUR 5,615 lower than for those treated with calcium carbonate (EUR 18,479 ± 17,228; 95% CI -10,066 to -1,164; p = 0.038).

CEA and Bootstrap Analysis

In the base case analysis, treatment with sevelamer resulted in greater LYs (0.06) and lower costs (EUR -5,615) per patient than treatment with calcium carbonate. The bootstrap analysis resulted in mean incremental LYs of 0.06 (95% CI 0.05-0.07) and cost savings of EUR 5,651 (95% CI -6,083 to -5,219) per patient (fig. 1). Notably, in 99.5% of the bootstrap samples, sevelamer was less costly than calcium carbonate. In 87% of the bootstrap samples, sevelamer was both less costly and more effective than calcium carbonate; in 0.05% of bootstrap samples, sevelamer was more costly and less effective than calcium carbonate, and in the remainder of the bootstrap samples (12.9%), sevelamer was either more costly and more effective than calcium carbonate or less costly and less effective than calcium carbonate.

Figure 2 presents the CEAC, plotted using the estimates of the probability that sevelamer was cost-effective over the range of WTP thresholds. The CEAC intersects the y-axis at approximately 99%, reflecting that nearly the entire joint density of incremental costs and effects involves cost savings for sevelamer versus calcium carbonate. The CEAC asymptotes to a value slightly less than 1 because not all of the joint density involves health gains.

A CEA was performed on the subgroup of patients from the INDEPENDENT-CKD study who did not initiate dialysis at any time during the 36-month duration of the study. The clinical and economic outcomes for this analysis are presented in table 3. Among patients who did not initiate dialysis during the study period, significantly fewer sevelamer-treated patients (n = 8, 10.5%) died compared to calcium carbonate-treated patients (n = 16, 25.4%; p = 0.025). The mean number of LYs per patient treated with sevelamer was 2.87 (95% CI 2.79-2.94), while the mean number of LYs per patient treated with calcium carbonate was 2.76 (95% CI 2.67-2.85), resulting in a difference of 0.11 LYs (95% CI -0.01 to 0.22; p = 0.008). The trends with respect to hospitalizations and costs were similar to those in the base case analysis.

The bootstrap analysis resulted in mean incremental LYs of 0.11 (95% CI 0.09-0.12) and cost savings of EUR 2,468 (95% CI -2,657 to -2,279) per patient (fig. 3). Similar to the base case analysis, sevelamer was less costly than calcium carbonate in 99.6% of the bootstrap samples. In 91.6% of the bootstrap samples, sevelamer was both less costly and more effective than calcium carbonate; in 0.1% of bootstrap samples, sevelamer was more costly and less effective than calcium carbonate, and in the remainder of the bootstrap samples (8.3%), sevelamer was either more costly and more effective than calcium carbonate or less costly and less effective than calcium carbonate. The CEAC reflects that nearly the entire joint density of incremental costs and effects involves cost savings for sevelamer versus calcium carbonate and the CEAC asymptotes to a value slightly less than 1 because not all of the joint density involves health gains.

The final total costs are shown in figure 4. The 107 patients of the sevelamer group spent in 36-month follow-up EUR 2,391,616.92 and the calcium group EUR 3,128,498.95 (difference EUR -736,882.03).

The recent INDEPENDENT-CKD study presented an opportunity to explore the cost-effectiveness of sevelamer versus calcium carbonate in patients with NDD-CKD. To the best of our knowledge, our study represents the first CEA in patients with NDD-CKD that is based on long-term, comparative outcomes data for survival and progression of CKD and the first economic evaluation of sevelamer in patients with NDD-CKD.

The base case analysis of this study established that sevelamer provided greater LYs and was less costly than calcium carbonate; cost savings associated with fewer hospitalizations for sevelamer offset the higher acquisition cost for sevelamer. In the bootstrap analysis, 87% of the samples showed that sevelamer was more effective and less costly than calcium carbonate. In the subgroup analysis that included only those patients that never initiated dialysis during the study, the results were more positive as sevelamer remained less costly and provided even greater LYs compared with calcium carbonate. In the bootstrap analysis for the subgroup population, 91.6% of the samples showed that sevelamer was more effective and less costly than calcium carbonate.

It is important to note that the estimate of cost savings associated with sevelamer may be conservative since reductions in hospital LOS were not monetized in the CEA. However, the results of the retrospective chart review of the patients included in the INDEPENDENT-CKD study showed that the average LOS per admitted patient was significantly lower for patients treated with sevelamer than for those patients treated with calcium carbonate (16.18 days for sevelamer vs. 22.10 days for calcium carbonate; 95% CI -10.5 to -1.30; p = 0.0123; not shown). This trend was also observed in the subgroup of patients that never initiated dialysis during the study (18.15 vs. 23.79 days, 95% CI -13 to 1.7; p = 0.13).

The fact that this economic evaluation used patient-level data collected in an RCT represents a significant strength of this study. To the degree that the INDEPENDENT-CKD study presents an unbiased estimate of all-cause mortality and disease progression in patients receiving either sevelamer or calcium carbonate for treatment of elevated serum phosphorus in patients with NDD-CKD, the results of our study represent the best available evidence of the cost-effectiveness of sevelamer versus calcium carbonate in patients similar to those included in the INDEPENDENT-CKD study. Access to patient-level data on outcomes and costs is particularly noteworthy in terms of analysis and internal validity.

The study, however, is not without limitations. Since the economic evaluation was based on an RCT, a natural limitation of the study concerns the generalizability of these results to real-world practice. However, attempts were made in the INDEPENDENT-CKD study to limit the invasiveness normally indicative of an RCT. For example, phosphate binders were given in an open-label fashion and clinicians were free to increase the initial dose of binder at their discretion to maintain serum phosphorus concentrations in the ranges suggested by KDOQI guidelines and according to the CKD stage of each patient. This would mitigate the risk of a lack of generalizability. Additionally, in the CEA, we did not include costs associated with protocol-driven outpatient physician visits or tests. Alternatively, these costs were excluded from the CEA under the assumption that these costs would not be different between treatment groups.

Since the INDEPENDENT-CKD study did not collect data on hospitalizations, a post hoc retrospective chart review of the patients included in the INDEPENDENT-CKD study was conducted to collect data regarding frequency of hospitalization, overall LOS, and discharging ward for the study duration. It would have been preferable that these data were collected at the time of the study as part of the study protocol. However, since complete data could be collected for all patients included in the study, we feel that any bias introduced as a consequence of the retrospective data collection was minimal.

Finally, the INDEPENDENT-CKD study did not collect data regarding patient quality of life or utility to allow the calculation of QALYs for the economic evaluation. In order to comply with the RCT and to minimize the use of external data and assumptions in the analysis, the ICER was reported as cost per LY gained only. As a consequence, comparison of results of this CEA with CEAs for other products or diseases for the purposes of allocating scarce resources across all health care is limited.

Despite these limitations, and until further research is conducted, this study currently represents the best estimate of the cost-effectiveness of sevelamer versus calcium carbonate for the treatment of elevated serum phosphorus in patients with NDD-CKD. From the perspective of patients, physicians, policy makers and budget holders, it is imperative to avoid hospitalization of patients whenever possible. The present analysis suggests that each euro spent on sevelamer instead of calcium carbonate will return EUR 1.80 to the NHS in hospitalization savings. Given the current global health environment characterized by escalating health care costs and cost minimization initiatives, sevelamer appears to be even more valuable.

Using patient-level data from the recently published INDEPENDENT-CKD study, this analysis demonstrates that sevelamer provides more LYs and is less costly than calcium carbonate in the treatment of hyperphosphatemia in patients with NDD-CKD in Italy.