|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to evaluate the cost-effectiveness of iron repletion using intravenous (i.v.) ferric carboxymaltose (FCM) in chronic heart failure (CHF) patients with iron deficiency with or without anaemia. Cost-effectiveness was studied from the perspective of the National Health Service in the UK.
A model-based cost-effectiveness analysis was used to compare iron repletion with FCM with no iron treatment. Using data from the FAIR-HF trial and publicly available sources and publications, per patient costs and clinical effectiveness of FCM were estimated compared with placebo. Cost assessment was based on study drug and administration costs, cost of CHF treatment, and hospital length of stay. The incremental cost-effectiveness ratio (ICER) of FCM use was expressed as cost per quality-adjusted life year (QALY) gained, and sensitivity analyses were performed on the base case. The time horizon of the analysis was 24 weeks. Mean QALYs were higher in the FCM arm (difference 0.037 QALYs; bootstrap-based 95% confidence interval 0.017–0.060). The ICER of FCM compared with placebo was €4414 per QALY gained for the FAIR-HF dosing regimen. Sensitivity analyses confirmed the base case result to be robust.
From the UK payers’ perspective, managing iron deficiency in CHF patients using i.v. FCM was cost-effective in this analysis. The base case ICER was clearly below the threshold of €22 200–€33 300/QALY gained (£20 000–£30 000) typically used by the UK National Institute for Health and Clinical Excellence and proved to be robust in sensitivity analysis. Improved symptoms and better quality of life contributed to this result.
Chronic heart failure (CHF) patients are often limited in their daily activities. Frequently reported symptoms are fatigue and dyspnoea, but also impaired physical working capacity, exhaustion, susceptibility to stress, and decreased mental and cognitive performance.1,2 Anaemia and iron deficiency are common findings in HF patients and may partially explain these symptoms. Anaemic HF patients are at risk of increased mortality, number of hospitalizations, and levels of natriuretic peptides, and reduced exercise capacity and impaired quality of life (QoL).3 Factors (e.g. renal dysfunction, haemodilution, and drugs) present in HF can contribute to the development of anaemia.3 A recent study found profound and general bone marrow dysfunction in CHF patients, another factor contributing to anaemia.4 Iron plays a key role in the uptake, transport, and storage of oxygen, which in CHF is often insufficiently supplied to the body. Iron deficiency in HF patients can exacerbate chronic diseases, and affects erythropoiesis and oxidative and immune mechanisms.3 In recent years, clinical research increasingly focused on iron therapy and administration of erythropoiesis-stimulating agents (ESAs) as treatment strategies for anaemia and iron deficiency in CHF patients.5 Intravenous (i.v.) iron repletion with ferric carboxymaltose (FCM) has been shown to improve symptoms and QoL in patients with CHF.6–8 FAIR-HF (see Appendix), a randomized, double-blind, placebo-controlled trial (n = 459) studied clinical and QoL benefits of treatment with FCM, an i.v. iron preparation, in iron-deficient CHF patients with New York Heart Association (NYHA) class II or III, a left ventricular ejection fraction of ≤40% (for patients in NYHA class II) or ≤45% (for patients in NYHA class III), a haemoglobin level at the screening visit of between 9.5 and 13.5 g/dL, and iron deficiency.9 Causes of HF in FAIR-HF patients were predominantly ischaemic [FCM n (%) 245 (80.6); placebo 123 (79.4)].6 Primary and secondary endpoints included NYHA functional class change from baseline and European Quality of Life-5 Dimensions (EQ-5D; EuroQol Group, Rotterdam, The Netherlands) questionnaire-based health-related quality of life (HRQoL).10 The FAIR-HF trial showed significantly better NYHA class changes from baseline in the FCM group compared with placebo, and FCM resulted in improved HRQoL [increased EQ-5D visual analogue scale (VAS) score change from baseline].6 This study evaluated the cost-effectiveness of iron repletion using i.v. FCM in CHF patients, from the perspective of the National Health Service (NHS) in the UK.
We performed a model-based cost-effectiveness analysis comparing a strategy of iron repletion with FCM with a standard strategy of no iron treatment in iron-deficient patients with or without anaemia. These strategies were generally equivalent to the strategies investigated in the FAIR-HF trial.9 Cost differences between the treatment strategies were also modelled and are reported in the Results section. A decision-tree model was used to facilitate sensitivity analysis. It allowed for performing deterministic as well as probabilistic sensitivity analyses.
The population basis for the clinical model inputs consisted of the 459 patients with NYHA functional class II and III at baseline, which formed the intention-to-treat (ITT) population of the FAIR-HF trial. Baseline characteristics are shown in Table 1. The FAIR-HF trial did not include British patients but was mostly performed in European countries (including Russia and Ukraine) with a predominantly Caucasian population. Therefore, the authors assume that clinical study results in a British population would not differ significantly from those of the actual trial.
In FAIR-HF, patients were randomly assigned to receive either FCM or placebo (normal saline). During a correction phase, patients received weekly injections until iron repletion was achieved.6 The total iron dose required for iron repletion was calculated at baseline using the Ganzoni formula.6 FCM was administered as an i.v. bolus injection of 4 mL equivalent to 200 mg of iron until repletion was achieved. Subsequently, during a maintenance phase, an injection was given every 4 weeks. Patients were assessed for NYHA functional class and QoL at baseline and Weeks 4, 12, and 24. QoL was represented by health state utilities measured with the EuroQol Group 5-Dimension Self-Report Questionnaire (EQ-5D). The EQ-5D is an instrument designed for self-completion by respondents. The instrument comprises two parts: first, respondents report their health status according to a five-dimensional classification (mobility, self-care, usual activities, pain/discomfort, anxiety/depression). Each dimension is represented by a three-level ordered category item, which leads to a total of 243 possible health ratings that are valued using the standard UK time trade-off value set. Secondly, the respondents record their self-perceived health status using a graduated visual analogue scale (VAS), with grades from 0 to 100. The EQ-5D is a generic QoL instrument which has been validated and shown to be sensitive, internally consistent, and reliable in the general population and different patient groups.11–13
In the cost-effectiveness analysis, available outcome measures for the different strategies assessed were (i) cost; (ii) effectiveness, expressed as the number of quality-adjusted life years (QALYs) accrued during the study period; and (iii) the incremental cost-effectiveness ratio (ICER), expressed as cost per QALY gained. Costs are reported in Euros (€) and Pounds sterling (£). For the conversion of £ to €, an exchange rate of 1.11 was used (mean exchange rate for 2009; source: www.oanda.com). Costs and benefits were not discounted given the short time horizon of the study period. Where necessary, cost information was inflation corrected using the ‘Unit Costs of Health and Social Care’ inflation indices of the Personal Social Services Research Unit (PSSRU).14 Costs borne by the patient and society as a whole were not incorporated, as they are not relevant from the NHS perspective.
The time horizon of the analysis was 24 weeks, corresponding to the duration of the FAIR-HF trial. Extrapolation of the time horizon to a longer term, e.g. lifelong time horizon, was not considered adequate as the FAIR-HF trial provides no information on longer term survival or other long-term effects. In the FAIR-HF trial, the rates of death, hospitalization, and serious or any adverse events were similar in the two study groups.6
Two approaches to cost estimation were used, as there was no detailed medical resource use information available that would have allowed direct assessment of direct medical costs. (i) In the base case analysis, only CHF-related hospitalization costs were taken into account. (ii) In a univariate sensitivity analysis, the costs of CHF treatment were estimated using cost proportions observed for patients in different NYHA classes.
In both approaches, UK FCM costs and UK FCM administration costs15,16 were additionally taken into account. Costs for adverse events were not included because there were not any clinically relevant differences between the FAIR-HF study arms. Table 3 provides an overview of the different approaches used in the analysis. All data derivations from the FAIR-HF raw data were performed by the authors and checked for consistency with the publication by Anker et al.6 where applicable.
The EQ-5D questionnaire (base case analysis) and VAS (sensitivity analysis) results from FAIR-HF,6 measured at baseline and at Weeks 4, 12, and 24, were converted into utility values as described above. QALYs were calculated by multiplying these utilities by the appropriate time periods for each individual. In order to achieve this, any changes in utility were assumed to occur in the middle of the periods defined by the assessment time points. Observations with missing values were imputed with the value of the last observation (last observation carried forward; LOCF), as was done in the FAIR-HF main clinical publication. Effectiveness was assessed as the number of QALYs accrued during the study. Clinical response to treatment, measured by change in NYHA class, was also assumed to be the same as in FAIR-HF.
In the base case analysis, medical resources taken into account were drug (FCM), FCM administration, and hospitalization for CHF (Table 2). Concerning drug usage, there was no wastage, as FCM vials were fully administered in each case. Hospitalization costs were determined by multiplying UK-based hospital length of stay for CHF patients17by the average number of hospitalizations seen in placebo patients in the FAIR-HF trial and, for patients in the FCM group, by the proportional length of stay in the FCM arm (average length of stay across all patients and hospitalizations in the FCM group divided by average length of stay in the placebo group). For patients remaining hospitalized after 24 weeks an artificial discharge date (baseline plus 24 weeks) was assumed.
In the second, NYHA class-based approach to cost estimation, CHF-related medical resource use (other than for FCM and FCM administration) was assumed to be represented by the typical cost of a patient falling into a given NYHA class: cost proportions observed for CHF patients in different NYHA classes published by Levin and Szucs18,19 were multiplied by published total costs for CHF patients20 and patient days per NYHA class according to data from the FAIR-HF study arms.
Costs for a hospital day for CHF patients were calculated using 2008–2009 NHS reference costs21 (Table 2). For FCM, the UK purchasing price (PP) was used in the base case analysis.22 Ambulatory administration costs were calculated using information on wages from the PSSRU data on unit costs for medical services.15 Costs of materials were taken into account according to information from the Falkirk & District Royal Infirmary.16
In order to assess the impact of statistical uncertainty around key model inputs, we performed a series of univariate and probabilistic sensitivity analyses.
In univariate sensitivity analysis, we varied (i) the mean duration of hospitalizations for CHF in the UK and (ii) the cost of a hospital day by ±30%; and (iii) drug costs by ±10% as no confidence intervals were available for these parameters. We further varied (iv) QALY difference; (v) proportional reduction of hospitalization days; and (vi) frequency of hospitalization in the placebo group on the basis of their confidence intervals. Further variations were: (vii) calculation of results considering only cases with complete data on utilities; (viii) calculation of costs via the NYHA class approach; and (ix) calculation of utilities using EQ-5D VAS scale scores.
Probabilistic sensitivity analysis (second-order Monte Carlo simulation; PSA) was based on distributions corresponding to the ranges of variation used in the univariate sensitivity analyses assessing the impact of parameter uncertainty (Table 2). PSA was based on 10 000 sets of randomly drawn input parameters.
The model was implemented and all Monte Carlo analyses were performed using TreeAge Pro 2011 Suite (TreeAge Software Inc., Williamstown, MA, USA). Further analyses were performed using Stata/IC 11 (StataCorp LP, College Station, TX, USA).
The clinical characteristics of the FAIR-HF ITT population are presented in Table 1. At baseline, there were 53 patients (17.4%) in NYHA functional class II and 251 (82.6%) in class III in the treatment group, and 29 patients (18.7%) in class II and 126 (81.3%) in class III in the placebo group.
In the base case analysis, mean QALYs were higher in the FCM arm (placebo, 0.298; FCM, 0.336); the difference was 0.037 QALYs [bootstrap-based 95% confidence interval (CI) 0.017–0.060] (cost and cost-effectiveness results are shown in Table 3). The ICER of FCM compared with placebo was €4414 (£3977) per QALY gained for the FAIR-HF dosing regimen. The FCM group yielded total costs of €852 (£768) and the placebo group of €687 (£619) over the study period. There were costs of €393 (£354) for the drug and €211 (£190) for administration compared with no costs on the placebo side. On the other hand, treatment with FCM saved €438 (£395) of costs for hospital treatment [FCM, €249 (£224); placebo, €687 (£619)], resulting in a net cost of the FCM strategy of €165 (£149) over 24 weeks.
Univariate sensitivity analysis was carried out to characterize the robustness of the base case results. Results ranged from dominance (i.e. being both cost saving and more effective) of the i.v. iron strategy [€1045 (£941) saved; effectiveness 0.037 QALYs] to €13 855 (£12 482) per QALY gained. Frequency and duration of hospitalization, QALY difference, and cost of a hospital day were the most influential parameters. Univariate sensitivity analysis results are presented in Table 3 and summarized in a Tornado diagram (Figure 1).
Analysis of mean days spent in each NYHA class, for FCM (placebo), resulted in 6.5 days (1) for NYHA class I, 58 days (40) for NYHA class II, 100 days (119) for NYHA class III, and <1 day (3) for NYHA class IV. The approach of assessing CHF treatment costs based on the time spent in each NYHA class resulted in costs of €1907 (£1718) for FCM and €1361 (£1226) for placebo, with a cost difference of €546 (£492) and a resulting ICER of €14 582 (£13 137), if the proportions according to Levin18 were used. If the distribution according to Szucs19 was used, costs were €1415 (£1275) for FCM and €887 (£799) for placebo, with a cost difference of €528 (£476) and a resulting ICER of €14 096 (£12 699). None of the parameters tested resulted in an ICER less favourable than €22 200–€33 300/QALY gained (£20 000–£30 000), the threshold usually regarded as acceptable by the UK National Institute for Health and Clinical Excellence (NICE).23
The PSA results showed mean costs of €877 (£790) for FCM and €686 (£618) for placebo [range: FCM, €608–€2050 (£548–£1847); placebo, €369–€1112 (£332–£1002)] and a mean effect of 0.038 QALYs (range –0.006 to 0.085). A total of 9866 scenarios (98.66%) were better than €22 200; 9968 scenarios (99.68%) were better than €33 300. A cost-effectiveness scatterplot and cost-effectiveness acceptability curve are shown in Figures Figures22 and 33, respectively. Generally, sensitivity analyses showed the results to be robust.
As the cost burden of CHF reaches up to 1–2% of healthcare budgets, cost-effective treatments to improve symptoms and QoL are needed.24 The cost-effectiveness of iron repletion using FCM, compared with no iron treatment, in CHF patients with iron deficiency derived from our trial-based analysis (€4414/QALY gained in the base case) was clearly below the cost-effectiveness threshold typically regarded as acceptable by NICE, of €22 200–€33 300/QALY gained.23 Mean QALYs were higher in the FCM arm, and improved symptoms and better quality of life contributed to economic benefits seen with FCM. Sensitivity analyses showed ranges of QALY differences (ICERs) from 0.017 to 0.060 (FCM dominant: €13 855) for the univariate and QALYs of –0.006 to 0.085 for the probabilistic analysis.
To the authors‘ knowledge FAIR-HF is the first study that compared FCM with placebo in a population of CHF patients with iron deficiency. Clinical implications of FCM use compared with other i.v. and oral iron compounds have been studied in various indications. Cost-effectiveness analyses of FCM use are scarce across indications, and none is available for CHF patients.
The use of ESAs in CHF patients is still under debate, and further research is needed in this regard.5 If ESA treatment was established as a therapeutic option, combination treatment with i.v. iron might have fewer cost implications in the long run than ESA treatment alone, as required ESA doses might be reduced.25–29
One important limitation of the present analysis is the lack of exhaustive medical resource use information; data on some uses of medical resources such as co-medications, devices (e.g. pacemakers), and ambulatory treatments could not be accounted for in the analysis, as they were not recorded in the FAIR-HF trial. As there is no indication of an increased use of unmeasured resource items in patients treated with i.v. iron vs. patients with no iron treatment, the authors do not expect that other resource items, had they been included in the analysis, would diminish the cost difference between FCM and placebo. The strategy to assess treatment costs via hospital days is likely to produce conservative results, because other potential resource savings, relating, for example, to outpatient visits, were not included in the calculation. In an alternative approach to cost assessment, treatment costs for CHF patients were estimated using the time spent in each NYHA class, and resulted in ICERs that are almost three-fold compared with the approach of using hospital days to estimate costs. Although the level of healthcare expenditure has been shown to be associated with NYHA class, this approach apparently cannot appropriately reflect changes in hospitalization rate and duration, as observed in FAIR-HF. Moreover, clinical treatment regimens most probably will not immediately follow and be adapted to changes in NYHA class (particularly improvements), preventing a rapid translation into cost changes. While it is of interest to study the distribution of CHF costs across NYHA classes, reverse use of such information to predict costs in other situations does not appear to be a suitable option. The study by Szucs et al.19 is based on a random sample of CHF outpatients in Switzerland, whereas the study by Levin et al. is based on the CHARM-trial,30 which contains data from CHF patients in 26 countries; both studies date back to 1999.
Some model inputs used in this analysis were subject to substantial uncertainty. It may, for example, be questionable whether NHS-reported Healthcare Resource Group (HRG) costs really reflect actual hospitalization costs. In the NYHA-based analysis, the most uncertain variable represented UK yearly costs for CHF treatment.
When discussing the generalizability of the presented results, it has to be noted that in routine clinical practice, fewer FCM administrations may be required, as FCM allows injections or infusions of up to 1000 mg in 15 min, which may lead to improved ICERs. The extent to which this dosing regime may be achieved remains uncertain. Our analysis was based on average hospitalization rates and lengths of stay. In routine practice, wide value ranges may occur and resource use may be structured differently, which could lead to an under- or overestimation of costs.
The effect of FCM was seen in a closely followed patient group with high NYHA classes( ~20% NYHA II and 80% NYHA III) and may be different in other collectives. Compared with a UK routine practice population, the FAIR-HF population was older, consisted of more females, and had lower diastolic and systolic blood pressure and higher co-morbidity.31 The impact of such differences is not currently predictable. The results of FAIR-HF stem from mostly pan-European study sites. There were no sites in the UK, but given the large European contribution one can assume that transferability of results to the UK is not substantially affected by geographical clinical variation. An analysis of anaemic vs. non-anaemic subgroups was not performed, as sample sizes would have been inadequate to perform a reliable analysis.
Study data cover a period of 24 weeks of treatment. Extrapolation of the time horizon to a longer term, e.g. a lifelong time horizon, was not considered adequate as the FAIR-HF trial provides no information on long-term survival or other long-term effects. The cost-effectiveness of longer term iron treatment remains unknown.
Given that the trial only collected very limited resource use data, the results do have an approximate character to some extent. In order to gain more in-depth knowledge about use of FCM in CHF patients, e.g. whether treatment and cost effects are sustained over a longer period, trials covering longer time periods and gathering further resource use should be the focus of further research. In conclusion, it can be noted that over the study period, treatment with FCM in iron-deficient CHF patients with or without anaemia improves symptoms and is likely to be cost-effective in routine clinical practice, from the perspective of the UK NHS.
Vifor Pharma Ltd, Switzerland. The authors accept full responsibility for the conduct and publication of the study and had access to the study data.
Conflict of interest: F.S.G. and P.R.B. report research funding from Vifor Pharma Ltd, Switzerland. M.S. reports speaker's honoraria and research funding from Vifor Pharma Ltd, Switzerland. P.G.B. and C.M. are employees of Vifor Pharma, Glattbrugg, Switzerland. C.M. holds shares of Galenica. T.D.S. reports no conflict of interest. P.P. reports receiving consulting fees from Vifor Pharma and Amgen, as well as honoraria and research support from Vifor Pharma. S.D.A. reports receiving consulting fees from Vifor Pharma, Amgen, Takeda, and Noxxon, honoraria for lectures from Vifor Pharma and Amgen, and research support from Vifor Pharma.
FAIR-HF: (clinicaltrials.gov: NCT00520780). Anker SD, Comin Colet J, Filippatos G, Willenheimer R, Dickstein K, Drexler H, et al. Ferric carboxymaltose in patients with heart failure and iron deficiency. N Engl J Med 2009;361:2436–2448.