Pharmacoeconomic analysis of paricalcitol and calcitriol in the treatment of secondary hyperparathyroidism in haemodialysis: impact of hospitalisations and survival

Summary

The objective of this study was to evaluate the cost effectiveness of paricalcitol injection compared with calcitriol injection when used to reduce parathyroid hormone levels in patients undergoing haemodialysis. A decision tree was developed to model the 1-year costs and outcomes of therapy for secondary hyperparathyroidism from a US government payer's perspective (2005 US$). Probabilities of hospitalisations and survival with paricalcitol and calcitriol were obtained from published observational studies.

When only drug costs and survival were considered, the incremental cost effectiveness of paricalcitol over calcitriol was $9,900 per life saved. When utilities were included, the incremental cost-effectiveness ratio for paricalcitol compared with calcitriol was $13,200 per quality-adjusted life year. When both drug and hospitalisation costs were included in a cost analysis, paricalcitol treatment was cost saving compared with calcitriol, and when hospitalisation costs were included in both the cost-effectiveness analysis and cost-utility analysis paricalcitol demonstrated first-order dominance, cost savings and cost effectiveness.

This decision analysis demonstrated that paricalcitol injection is both cost effective and cost saving compared with calcitriol injection.

Keywords: :

• chronic renal failure
• hyperparathyroidism
• cost effectiveness

Introduction

Secondary hyperparathyroidism (SHPT) is present in more than 50% of haemodialysis (HD) patients with stage 5 chronic kidney disease (CKD) (also known as end-stage renal disease (ESRD))1 and is the result of low serum calcitriol and calcium levels, and high serum phosphorus levels2–4. Common sequelae associated with SHPT include renal osteodystrophy, pain related to loss of skeletal structure, fracture5,6 and vascular complications7–9.

For the past two decades, the vitamin D receptor (VDR) activator calcitriol, which acts directly on the parathyroid gland to reduce parathyroid hormone (PTH) synthesis and increase calcium absorption in the gut10, has been used to control elevated PTH. Whilst effective in controlling SHPT, calcitriol complicates therapy by increasing serum levels of calcium, phosphorus and calcium–phosphorus product (Ca × P)11. Calcitriol-related toxicities have been implicated in elevating the risk of extraskeletal calcification in soft tissue, periarticular areas, vasculature and heart valves7,10,12. To avoid some of the untoward effects experienced with calcitriol therapy, VDR activators have been developed by modifying the chemical structure of the calcitriol molecule. Slight structural changes can affect the way a VDR activator interacts with vitamin D binding protein and VDRs. This may thereby enable the VDR activator to exert selective actions on target tissues.

Paricalcitol is a third-generation VDR activator that lacks the exocyclic carbon at position 19 and acts as a selective VDR activator13,14. Pivotal randomised controlled trials conducted in patients with stage 5 CKD demonstrated a significant decrease in serum-intact PTH and serum alkaline phosphatase, and lower Ca × P compared with placebo15–17. Paricalcitol selectively upregulates the VDR in the parathyroid glands without increasing VDRs in the intestine and is less active on bone resorption compared with calcitriol14,17–19.

In addition to new data suggesting a 20% survival benefit associated with the use of activated injectable VDR activators in HD patients with SHPT20, recent studies comparing paricalcitol with calcitriol indicate that patients who receive paricalcitol while undergoing long-term HD have fewer hospitalisations per year, fewer hospital-days per year and a significant survival advantage over those who receive calcitriol21,22. Injectable paricalcitol-treated patients also achieve reduced PTH significantly faster than injectable calcitriol-treated patients23. Paricalcitol appears to be effective at either a 4:1 dose ratio (paricalcitol to calcitriol), as recommended by earlier studies, or at a 3:1 dose ratio demonstrated in a more recent studies24,25.

To date, no published cost-effectiveness studies have compared intravenous (iv) paricalcitol with iv calcitriol. Using a decision-analysis model, the authors sought to compare the cost effectiveness of these agents for the treatment of SHPT in HD patients with stage 5 CKD by incorporating the probabilities of hospitalisation and survival, costs related to drug therapy and hospitalisation, and patient utilities.

Methods

Decision model

A decision model (Figure 1) was used to assess the annual drug and hospitalisation costs and consequences of treating SHPT in HD patients with stage 5 CKD, with either paricalcitol injection or calcitriol injection, with a 1-year time horizon from the US government payer's perspective. TreeAge software DATA 4.0 and Microsoft Excel 2000 were used to design the decision tree and to conduct the analysis. This model considered hospitalisations and 1-year survival after the decision to prescribe either regimen. Probabilities of each outcome and costs associated with therapy and hospitalisation were obtained from recently published data21,22.

Figure 1. Decision tree representing the primary outcomes (hospitalisation and survival) that occur for patients with secondary hyperparathyroidism treated with either intravenous paricalcitol or calcitriol.

Costs used in the model

Costs included in the model are shown in Table 1 and consisted of the projected yearly reimbursement for injectable paricalcitol and calcitriol as well as the diagnosis-related group (DRG) reimbursement rate for hospitalisation. All costs were expressed in 2005 $US. No discounting was performed because the time frame of the model was 1 year. Costs remain the same for all three analyses, except for the threshold analysis involving costs.

Table 1. Drug and hospital costs included in the analyses.

Resource item	Cost ($)
Paricalcitol iv (cost per μg)*	4.00
Calcitriol iv (cost per μg)^†	9.60
Weighted cost per hospitalisation^‡	6,434
Estimated hospital cost per day^‡	1,038

iv, intravenous.

* Value based on 2005 Medicare reimbursement fees (note: cost per μg, dosing ratio used in this study was 3:1 paricalcitol:calcitriol); J2501 Injection Paricalcitol 1 μg of $4.00²⁶

^† Value based on 2005 Medicare reimbursement fees; J0636 Injection Calcitriol 0.1μg of $0.96, therefore, 1 μg equals $9.60²⁶

^‡ Value based on diagnosis-related group costs with haemodialysis procedure, weighted by number of dialysis procedures; source was the 2006 DRG Handbook (using 2004 data, adjusted upward by 4.2% to convert to 2005 US$)²⁷

The mean cost for each drug was the 2005 Medicare reimbursement rate26. The dosage for calcitriol used to determine this cost was 1 μg administered three times per week and the dosage for paricalcitol was 3 μg three times per week. Although early studies of paricalcitol dosing suggested an initial 4:1 dosing ratio relative to calcitriol to achieve suppression of PTH, several more recent studies suggested acceptable clinical efficacy at a 3:1 ratio22–24,28. One clinical trial compared calcitriol with paricalcitol in a randomised, multicentre, blinded fashion23, and during the maintenance phase of that study the ratio of paricalcitol to calcitriol ranged from 2.3:1 to 3.0:1, with a calculated mean weighted dose ratio of 3.14:1 at 12 months28. Furthermore, the survival study by Teng compared these two agents and found the mean dosing ratio of paricalcitol to calcitriol at 12 months to be 3.3:122. Therefore, a ratio of 3:1 was used to determine the dosing regimens a priori. Since calcitriol-treated patients spend 9.17 more days in the hospital compared to paricalcitol-treated patients, calcitriol-treated patients would receive six doses of calcitriol in the hospital, where as paricalcitol-treated patients would only receive three doses. Therefore, assuming three dialysis sessions per week, the number of doses administered within the dialysis center per year was 150 for calcitriol and 153 for paricalcitol.The final drug cost per year was calculated by multiplying the VDR activator dose (μg) by cost per μg, multiplied by either 150 doses per year for calcitriol or 153 doses per year for paricalcitol. Costs associated with adverse events related to medications were not included in the model.

A mean value from DRG Medicare reimbursement rates for hospitalisations with HD ranked in the top four procedures were used to represent hospitalisation costs in the model27. This mean was weighted based on the frequency of HD procedures for the following DRGs: DRG 127 Heart Failure & Shock; DRG 361 Renal Failure; DRG 144 Other Circulatory System Diagnosis with complication or co-morbidity (CC); DRG 296 Nutritional & Metabolic Disorder Age >17 with CC; DRG 416 Septicemia Age >17; DRG 331 Other Kidney & Urinary Tract Diagnoses Age >17 with CC; DRG 138 Cardiac Arrhythmia and Conduction Disorder with CC; DRG 294 Diabetes >35; DRG 132 Arteriosclerosis with CC; DRG 087 Pulmonary Edema & Respiratory Failure; DRG 463 Signs & Symptoms with CC; DRG 089 Simple Pneumonia & Pleurisy Age >17 with CC; and DRG 236 Fractures of Hip & Pelvis. The following DRGs were eliminated from the analysis due to lack of potential impact from SHPT: DRG 018 Cranial & Peripheral Nerve Disorders with CC; DRG 096 Bronchitis & Asthma Age >17 with CC; and DRG 429 Organic Disturbance & Mental Retardation. Because the Medicare reimbursement rate used was from 2004, a 4.2% (medical component of consumer price index29) adjustment was applied to convert to 2005. The mean weighted cost per hospitalisation for these DRGs adjusted to 2005 was $6,434, and the estimated cost per hospital day was $1,038. This is a conservative estimate considering the mean hospitalisation charge for a principal diagnosis of SHPT using ICD-9 diagnosis code 588.81 from the 2004 Healthcare Costs and Utilization Project is $36,306 and the mean length of stay is 5.7 days. Assuming a 2:1 cost-to-charge ratio and a 4.2% discount rate, the 2005 cost per hospitalisation would be $18,915 and the cost per hospital day would be $3,31830. The reimbursement perspective may explain the difference in this cost. The first method used Medicare reimbursement rates, whereas the second method was from all payers. Hospitalisation costs were analysed using two approaches, cost per hospitalisation and cost per hospital day.

Probabilities used in the model

The probabilities of patients in each treatment option requiring hospitalisation, and subsequent survival or death, were obtained from previously published literature and included in the model. A Medline search was conducted for the period January 1960 through to June 2006 (inclusive). The search revealed no randomised controlled trials of paricalcitol injection and calcitriol injection with outcomes data for hospitalisation or survival. However, the search identified two observational studies comparing paricalcitol injection and calcitriol injection with respect to the incidence of hospitalisation and survival. The results of these studies were used to calculate the probabilities of these events.

Dobrez et al21 examined utilisation data derived from a large, national dialysis provider patient cohort, which included 11,442 patients receiving a minimum of 60 days of HD and at least 10 doses of VDR activator therapy. Descriptive (crude) results showed 59.6% (95% confidence interval (CI) 58.2–61) of patients who started on paricalcitol injection had at least 1 hospitalisation, a mean of 2.4 hospitalisations per year for all patients and a mean of 17.2 hospital days per year, compared with 75.2% (95% CI 74.3–76.2) with at least 1 hospitalisation, a mean of 2.61 hospitalisations per year for all patients and a mean of 19.8 hospital days per year with calcitriol injection. Multivariate results, adjusting for potential confounders between the cohorts, demonstrated that patients who remained on paricalcitol experienced 0.846 (95% CI 0.687–1.006) fewer all-cause hospitalisations per year and 9.17 (95% CI 5.56–12.79) fewer all-cause hospital days compared with patients who remained on calcitriol (p<0.001). Potential confounders included were days of HD prior to initiating VDR activator therapy, and baseline PTH, serum calcium, serum phosphorus, serum albumin, age, gender, ethnicity and comorbidities. Since the multivariate result imparts a higher level of evidence compared with descriptive results, multivariate results are used in the present model. To incorporate the multivariate results in the model, hospitalisations are included as a cost savings with paricalcitol. This is calculated by multiplying the average reimbursement per admission ($6,434) by 0.846 fewer hospitalisations per year, resulting in a $5,443 savings with paricalcitol therapy. The overall net cost of paricalcitol is the annual drug cost of paricalcitol ($1,836) minus the savings due to fewer hospitalisations of $5,443, resulting in net savings of $3,607.

Teng et al22 examined historical cohort data (n=67,399) to compare survival rates among patients undergoing HD who were started on paricalcitol or calcitriol. Differences in survival were apparent at 12 months and continued to increase with time (p<0.001). Subsequently, the authors reported a calculated survival of 0.84 and 0.80 at 1 year for paricalcitol and calcitriol, respectively31. These survival rates were used in the model in this study. No differences in survival between patients requiring hospitalisation versus those not requiring hospitalisation were assumed.

Utilities used in the model

As described below in the analysis section, different types of analyses were conducted in this study, including a cost-utility analysis. For this analysis, patient-based utilities were used to calculate quality-adjusted life years (QALYs). Utilities are values that represent the strength of an individual's preference for different health outcomes under conditions of uncertainty. The conventional utility scale has a utility of 0.0 for dead and a utility of 1.0 for complete health32. From the utility values for each health outcome, QALYs can be determined by multiplying the utility score by the probability for each health outcome life years gained as a result of treatment, as shown in Table 2.

Table 2. Calculation of QALYs.

	Hospitalisation probability	Survival probability	Utility score	QALY*
Paricalcitol
Pathway 1: hospitalised, alive	0.596	0.84	0.42	0.210
Pathway 2: hospitalised, die	0.596	0.16	0	0
Pathway 3: not hospitalised, alive	0.404	0.84	0.51	0.173
Pathway 4: not hospitalised, die	0.404	0.16	0	0
				Total QALYs for paricalcitol = 0.383
Calcitriol
Pathway 5: hospitalised, alive	0.752	0.80	0.42	0.253
Pathway 6: hospitalised, die	0.752	0.20	0	0
Pathway 7: not hospitalised, alive	0.248	0.80	0.51	0.101
Pathway 8: not hospitalised, die	0.248	0.20	0	0
				Total QALYs for calcitriol = 0.354

QALY, quality-adjusted life year.

* QALYs were calculated by multiplying the probability of hospitalisation by survival probability and then by the utility score for each pathway. Paricalcitol treatment was calculated to have an increase of 0.03 QALYs compared with calcitriol.

There are no published data on measured utilities for either hospitalised or non-hospitalised HD stage 5 CKD patients or ESRD patients with SHPT. However, utility scores for stage 5 CKD and dialysis in ambulatory patients have been reported to range from 0.404 to 0.6133–37. The value 0.51 was chosen as the utility score to calculate QALYs for non-hospitalised stage 5 CKD patients with SHPT because it represents the midpoint of the published range of values available for this patient population. The same utility value was used for both paricalcitol and calcitriol, with only life years gained being different between the two options.

To determine QALYs in hospitalised HD patients with stage 5 CKD with SHPT, the incremental decrease in the utility score was considered above what would occur as a result of hospitalisation. As congestive heart failure is a common diagnosis and reason for hospitalisation among HD patients38–41, heart failure utilities obtained both from hospitalised and non-hospitalised heart failure patients were compared. In a study by Capomolla et al42, the utilities for hospitalised and non-hospitalised patients with heart failure were 0.63 (± 0.22 sd) and 0.72 (± 0.17 sd), respectively (p<0.008), representing a 0.09 decrease in utility for those hospitalised. Therefore, for the hospitalised stage 5 CKD HD patients with SHPT in this analysis, a conservative estimated decrease of 0.09 (± 0.05 sd) was used in the calculation of QALYs.

Analysis

Three different types of economic analyses were used in the study: cost analysis (where the two options were compared based on cost only); cost-effectiveness analysis (where the two options were compared based both on cost and years of life gained); and cost-utility analysis (where the two options were compared based on costs and QALYs gained).

First, a cost analysis was conducted to compare the treatment options under two scenarios: (1) based on drug costs alone; and (2) based on drug costs and hospitalisation costs. In the second scenario, hospitalisation costs were considered in two ways: (1) based on cost per hospitalisation (i.e. as is the case under a DRG); and (2) based on cost per hospital day (e.g. as is the case under a per-diem payment system).

Next, a cost-effectiveness analysis was conducted with survival (lives saved) as the primary endpoint of interest. Again, costs were considered as being either: (1) drug costs only; or (2) drug costs plus hospitalisation-related costs, with hospitalisation costs considered as either cost per hospitalisation or cost per hospital day.

The third type of analysis performed was a cost-utility analysis, in which the patient-based utilities previously discussed were used, together with survival, to determine the QALYs for each option. The two treatment options (iv calcitriol and iv paricalcitol) were then compared based on cost per QALY. Again, this analysis was conducted considering the different iterations of cost used in the first two analyses.

Finally, to determine the robustness of the findings over a range of different probabilities and costs, a series of sensitivity analyses were conducted. These sensitivity analyses were designed to determine the thresholds or breakeven points (points at which the decision to use one drug over the other would change) for the various inputs (costs and probabilities) used in the model. In this analysis, hospitalisation costs, hospitalisation rate, paricalcitol cost and dose were included to determine at which point paricalcitol was cost equivalent compared with calcitriol. In addition, a sensitivity analysis was performed varying the utilities for the stage 5 CKD HD patients with SHPT who were not hospitalised within the range of values derived from the literature for stage 5 CKD and dialysis patients.

Results

Results of the different analyses performed in the study are summarised in Table 3a and b, and are described in each of the sections below.

Table 3a. Results of CA.

Type of analysis	Treatment options	Costs included	Cost ($)*†	Incremental cost ($)*†
CA	Paricalcitol iv	Drug only	1,836	396
	Calcitriol iv	Drug only	1,440
	Paricalcitol iv	Drug and hospitalisation	(3,607)	(5,047)
	Calcitriol iv	Drug and hospitalisation	1,440
	Paricalcitol iv	Drug and hospitalisation per diem	(7682)	(9,122)
	Calcitriol iv	Drug and hospitalisation per diem	1,440

iv, intravenous; CA, cost analysis.

* Negative dollar values (savings) are shown in parentheses.

^† Negative dollar values appear as only the hospitalisation savings associated with paricalcitol were applied to this analysis.

Table 3b. Results of CEA and CUA.

Type of analysis	Treatment options	Costs included	Effectiveness*	Incremental effectiveness	CE ratio
CEA	Paricalcitol iv	Drug only	0.84	0.04	$9,900/life saved
	Calcitriol iv	Drug only	0.80
	Paricalcitol iv	Drug and hospitalisation	0.84	0.04	Dominance^†
	Calcitriol iv	Drug and hospitalisation	0.80
	Paricalcitol iv	Drug and hospitalisation per diem	0.84	0.04	Dominance^†
	Calcitriol iv	Drug and hospitalisation per diem	0.80
CUA	Paricalcitol iv	Drug only	0.378	0.03	$13,200/QALYs
	Calcitriol iv	Drug only	0.348
	Paricalcitol iv	Drug and hospitalisation	0.378	0.03	Dominance^†
	Calcitriol iv	Drug and hospitalisation	0.348
	Paricalcitol iv	Drug and hospitalisation per diem	0.378	0.03	Dominance^†
	Calcitriol iv	Drug and hospitalisation per diem	0.348

iv, intravenous; CE ratio, incremental cost divided by incremental effectiveness; CEA, cost-effectiveness analysis; CUA, cost-utility analysis; QALY, quality-adjusted life year.

* Values in this column are survival (in years) for CEA and are QALYs for CUA.

^† ‘Dominance’ means that paricalcitol is both cost savings and more effective than calcitriol.

Cost analysis

Mean yearly treatment (drug only) costs associated with paricalcitol and calcitriol were $1,836 and $1,440, respectively. This represents an incremental (higher) cost of $396 per year for paricalcitol. However, when the probabilities and costs of hospitalisation were added to the model, paricalcitol treatment resulted in an overall cost saving of $5,047 compared with calcitriol. Based on cost per hospitalisation of $6,434, and assuming 0.846 fewer hospitalisations per patient year for paricalcitol, the incremental difference (total hospitalisation cost savings) between paricalcitol and calcitriol treatment was $5,443. In other words, the reduced hospitalisations associated with paricalcitol treatment resulted in an overall reduction in healthcare costs compared with calcitriol, despite the slightly higher drug cost. The savings were even greater ($9,122) when hospital costs were based on a cost per hospital day (and assuming 9.17 fewer hospital days).

Cost-effectiveness analysis

Based on drug costs only, and assuming 1-year survival rates of 0.84 and 0.80 for paricalcitol and calcitriol, respectively, the incremental cost effectiveness of paricalcitol over calcitriol was $9,900 per life saved. When costs of hospitalisation were added to the model, paricalcitol demonstrated dominance; that is, not only did paricalcitol treatment result in lower overall costs but it was also more effective. In pharmacoeconomic terms, this is referred to as ‘first-order dominance’. The per-day hospital cost differential was also examined in this manner and was found to result in even greater first-order dominance.

Cost-utility analysis

Results of the cost-utility analysis are shown in Tables 2 and 3. For the cost-utility analysis, QALYs were calculated for each pathway of the decision tree. The paricalcitol and calcitriol QALY scores were 0.378 and 0.348, respectively, a difference of 0.030. When considering drug cost only, the incremental cost per QALY for paricalcitol compared with calcitriol was $13,200. When hospitalisation costs were added to the model, paricalcitol treatment again demonstrated first-order dominance (cost saving and more effective) over calcitriol.

Threshold and sensitivity analysis

A series of one-way sensitivity analyses were performed to determine the values for cost per hospitalisation, number of hospitalisations, number of inpatients days, paricalcitol cost, and paricalcitol dose that would result in the preferred treatment option switching from paricalcitol to calcitriol for cost-effectiveness analysis. In the scenario that included both drug and DRG-based hospitalisation costs in the sensitivity analysis, paricalcitol remained cost saving until: (1) the cost per hospitalisation (DRG reimbursement rate) was less than $468, as shown in Figure 2; (2) the difference in paricalcitol hospitalisation rate was less than 0.0611 hospitalisations per year, as shown in Figure 3; (3) the annual cost of paricalcitol was greater than $6,885; and (4) the average annual dose ratio of paricalcitol to calcitriol was 11.2:1 μg, as shown in Figure 4. Results were similar when costs per hospital day were substituted for cost per hospitalisation. Since none of these threshold values are likely to occur in practice, the decision to use paricalcitol over calcitriol seems to be very robust.

Figure 2. Sensitivity analysis varying cost per hospitalisation.

Figure 3. Sensitivity analysis varying difference in hospitalisations.

Figure 4. Sensitivity analysis varying paricalcitol dose.

The utility score that was used in the cost-utility analysis for stage 5 CKD HD non-hospitalised patients with SHPT was selected based on the range of published scores for ESRD and dialysis patients, and the typical signs and symptoms associated with SHPT. Owing to the uncertainty between the estimated utility value and the true utility value, a sensitivity analysis was performed by varying the score used to calculate QALYs in the model based on the utility standard deviation. The subsequent range of differences in QALYs was 0.024–0.035. The incremental cost per QALY, considering drug cost only, varied between $11,314/QALYs to $16,500/QALYs. When hospitalisation costs were included in the model, paricalcitol remained dominant over calcitriol.

Discussion

Paricalcitol has emerged as the standard of care for the treatment of SHPT in HD patients with stage 5 CKD in the US owing to its efficacy and minimal impact on serum calcium and phosphorus. Compared with calcitriol, paricalcitol is associated with fewer hospitalisations22 and improved survival21. However, if one considers only the cost of the drugs, paricalcitol is more expensive than calcitriol. This analysis was conducted by taking into consideration reimbursement costs related to these improved outcomes, in addition to the reimbursement cost of the drug, and thereby represents the perspective of the payers.

This analysis demonstrated that paricalcitol is less costly compared with calcitriol when considering both drug costs and costs associated with hospitalisation. Furthermore, paricalcitol is both cost saving and more effective (first-order dominance) when costs and either survival or QALYs are included in the analysis.

Recently, two additional observational studies of the impact of VDR activators on mortality have been published. The first study examined 5,299 patients on paricalcitol and calcitriol from 1999 through to 2004. In unadjusted models, mortality was lower in patients on paricalcitol (0.79, 0.68–0.92) versus calcitriol, p<0.0543. However, in adjusted models, the mortality difference was not statistically significant. Since the sample size of this study was significantly smaller than the Teng study, it may not have been powered sufficiently to assess non-inferiority. The second study, by Kalantar-Zadeh, examined associations between survival, quarterly laboratory values and administration of paricalcitol in a 2-year cohort of 58,058 maintenance HD patients44. Administration of any dose of paricalcitol was associated with improved survival in time-varying models. The benefit of time-varying models allows for the accountability of variations in laboratory values and paricalcitol dose over time. While these studies provide additional insight, they do not merit changing the assumptions for this model.

Limitations

This study has a number of limitations, the first of which is the ratio of the dose of paricalcitol to the dose of calcitriol (3:1) used in this model, which was based on an extrapolated ratio from a clinical trial23. However, even if the higher dosing ratio (4:1) had been used in this analysis, the results would not have changed. Based on the threshold analysis, the dosing ratio would have to be greater than 11:1 before calcitriol would become the preferred treatment option.

Another limitation is that the hospitalisation utilisation and survival data used in this model were drawn from a quasi-experimental observational study. Data from observational studies are sufficient for generating hypotheses but not for determining cause and effect relationships. However, since no randomised controlled trials have been conducted that compare these agents with respect to hospital admissions and inpatient days, observational data are the current best evidence for these outcomes. On the other hand, there is controversy about using data from controlled trials because such studies are highly controlled and thus often not representative of care that occurs in ‘real world’ medical practice45. Nevertheless, in the threshold analysis conducted here, paricalcitol remained the preferred option until the difference in hospitalisations was less than 0.0611 per year, an unlikely value given the current evidence in the literature showing that paricalcitol injection was associated with a descriptive mean difference of 0.21 hospitalisations, and a 0.846 difference when multivariate analysis, which adjusted for confounding variables, were compared with calcitriol injection.

A third limitation of this analysis was that the probabilities or costs of adverse events of the two treatment options were not included. Adverse events reported with a frequency of ≥5% for paricalcitol include nausea, vomiting and oedema, whilst for calcitriol the most common side effects are vitamin D toxicity and hypercalcaemia. Data from a randomised controlled trial indicate that calcitriol-treated patients demonstrated a nearly 40% relative increase in hypercalcaemia compared with paricalcitol (9.2 vs. 6.7%, respectively)23. However, any cost and additional resource utilisation of this toxicity difference beyond that which requires hospitalisation was not included in the model due to lack of any published data.

Another potential limitation is that the utilities used in the cost-utility analysis were specific to stage 5 CKD or dialysis patients, which did not differentiate between patients with or without SHPT. It is estimated that 40–60% of stage 5 CKD dialysis patients are diagnosed with SHPT. To assess the importance of this limitation, a sensitivity analysis was conducted varying the utilities within the published ranges. The impact of this analysis did not influence the decision, thus maintaining the robustness of the results. Since the difference in hospitalisation utilities was based on a surrogate for SHPT hospitalisations, a sensitivity analysis eliminating any difference in utilities for hospitalisations was performed. Paricalcitol injection remained dominant when both drug and hospitalisation costs were included in the analysis. Furthermore, while a variety of sensitivity analyses were conducted, they were done so in a one-way fashion rather than changing multiple variables simultaneously. However, given the results of this threshold analyses it is unlikely that further analyses would alter the findings.

Although patients who are hospitalised may be at greater risk of death, the probabilities for survival and hospitalisations were from two independent studies. Assuming no difference in survival among hospitalised and non-hospitalised patients is a conservative approach favouring calcitriol therapy.

Management of SHPT starting in stage 3 CKD is important in controlling downstream complications. Clinical thought leaders and recent guidelines by the National Kidney Foundation confirm this by recommending early management of SHPT to prevent bone and cardiovascular complications, to preserve normal calcium and phosphate homeostasis, and to correct vitamin D deficiency. Payers must address, and balance the needs and issues of all stakeholders, including the improvement of clinical outcomes, maintenance of quality and access to care, improvements in economic efficiency, management of reimbursement and utilisation, and improvements in compliance and persistence.

Conclusion

When considering reductions in resource consumption for data obtained from observational studies, specifically hospitalisation or hospital days, this analysis suggests that paricalcitol is cost saving compared with calcitriol for the treatment of SHPT in HD patients, from a payer's perspective. Furthermore, paricalcitol is more cost effective than calcitriol when considering drug cost and survival or QALYs. Physicians, government and managed care payers, particularly integrated health systems, should consider these findings when making decisions regarding the use of these two agents. Long-term studies that assess the impact of parathyroid suppression in this sensitive patient group will greatly aid in improving the assumptions made in this cost-effectiveness model. Additional comparative studies are necessary to validate these results.

Acknowledgements

With respect to financial interests of the authors, Steven E Marx, Joel Melnick, Raimund Sterz and Laura A Williams are employees of Abbott Laboratories. Abbott Laboratories is the manufacturer of both of the products compared in this analysis. Glen T Schumock and Jose AL Arruda have served as consultants to Abbott Laboratories.

Notes