Bazedoxifene is a novel selective estrogen receptor modulator (SERM) for the prevention and treatment of osteoporosis.1, 2 In a pivotal 3-year, phase 3 trial, bazedoxifene significantly reduced the risk of new vertebral fractures versus placebo in postmenopausal women with osteoporosis.2 In a subgroup of women at higher risk of fracture, bazedoxifene significantly reduced nonvertebral fractures compared with placebo and raloxifene.2 Bazedoxifene treatment was generally well tolerated and showed a favorable safety profile.2 Based on these findings, bazedoxifene was granted marketing authorization in the European Union from the European Medicines Agency in February 2009.3
In addition to efficacy and safety data, evaluation of the cost-effectiveness is becoming an important consideration in reimbursement decisions of new drugs.4 The cost-effectiveness of bazedoxifene was demonstrated compared with placebo in several European countries.5, 6 Other osteoporotic agents (eg, alendronate, denosumab, raloxifene, risedronate, strontium ranelate,) have also been shown to be cost-effective compared with placebo.7–9 For decision makers, it would be important to know whether bazedoxifene represents a good value for the money compared with relevant active comparators. Health economic evaluations should ideally compare a new intervention with the interventions it is likely to replace. This study was therefore designed to estimate the cost-effectiveness of bazedoxifene compared with another SERM, raloxifene, in the treatment of postmenopausal osteoporotic women. Importantly, this was the first cost-effectiveness analysis between active comparators in osteoporosis that used efficacy data from a single trial.
Materials and Methods
The cost-effectiveness of treatment for 3 years with bazedoxifene was compared with raloxifene using a Markov microsimulation model. The economic model has been validated10 and extensively used to assess the cost-effectiveness of osteoporosis treatments11–14 and screening,15 and to evaluate the economic implications of poor adherence with osteoporosis medications.16–18 The study population included women aged over 60 years with osteoporosis (ie, bone mineral density [BMD] T-score ≤ −2.5 and/or prevalent vertebral fractures). Analyses were conducted in a Belgian setting using a healthcare payer perspective as recommended by the methodological guidelines for pharmacoeconomic evaluations in the country.19
An updated version of the model using a 6-month cycle length was used in this analysis.14 The model was programmed using the software TreeAge Pro 2011 (TreeAge Pro Inc., Williamston, MA, USA). The Markov model health states are “no fracture,” “death,” “hip fracture,” “clinical vertebral fracture,” “wrist fracture,” “other fracture,” “venous thrombotic events” (VTE), and the corresponding postfracture states. All the patients, one at a time, began in the “no fracture” state and had, every 6 months, a probability of having a fracture at the hip, clinical vertebrae, wrist, or other site, of having a VTE, or of dying. Patients in a fracture state can stay in the same fracture state if they refracture, change to another fracture state, die, or change in the next cycle to the postfracture state. It is also possible for patients to get VTE after all fracture heath states. Postfracture states were created as some parameters (eg, fracture disutility) were only estimated over a 1-year period. Patients being in any postfracture state might have a new fracture (all fracture types are possible), VTE, die, or move to the “no fracture” state. Prior fractures were recorded by tracker variables and used in calculations of transition probabilities, costs, and utilities in order to reflect the long-term effects of fractures. To capture the extraskeletal benefits of raloxifene, breast cancer was included, in a separate analysis, as a possible event using the same methodology as for fracture states.
A description of the different components of the model is provided below. Fracture data are included in Table 1. Please also refer to previously published research for further details and limitations of the model.10
Table 1. Incidence of Fractures, Costs, and Utilities Used in the Model
For normal distributions, an SD of 15% of the mean was assumed. Parameters of other distributions were derived from the 95% confidence intervals.
y = years of age; CV = clinical vertebral.
Incidence (annual rate per 1000) of first fracture
Hip, extra costs in the year following the fracture
Hip, yearly long-term costs
CV, first 6 months
Wrist, first 6 months
Other, first 6 months
Health state utility values
0.827 (60–69 y), 0.762 (70–79 y), 0.711 (+80 y)
Hip (first year/subsequent years)
0.80 (0.770–0.825)/0.90 (0.885–0.910)
CV (first year/subsequent years)
0.72 (0.660–0.775)/0.93 (0.916–0.946)
Wrist (first year/subsequent years)
Other (first year/subsequent years)
The base-case population was women aged 70 years with a BMD T-score below the threshold value for osteoporosis (ie, BMD T-score ≤ −2.5) and without any fracture. The incidence of first hip fracture in the general women population was derived from the national database of hospital bills (average of the years 2005–2007).20 Because the incidence of other fractures was not known, we assumed that the age-specific ratio of index fracture to hip fracture in Belgium was the same as found in Sweden.21 This assumption, used in the development of many FRAX models including Belgium,22 appears to hold true for West European countries, the United States, and Australia.23
The incidence of fracture in the general population was further adjusted to accurately reflect the fracture risk in women with low BMD. The relative risk for BMD was calculated using a method as described.23 This method estimates the risk of individuals below a threshold value in comparison with that in the general population. BMD values at the femoral neck were derived from the National Health and Nutrition Examination Survey (NHANES) III24 database, and a 1 SD decrease in BMD was associated with an increase in age-adjusted relative risk of 1.8, 1.4, and 1.6 for clinical vertebral, wrist, and other fracture, respectively.25 For hip fracture, the age-adjusted relative risk ranged from 3.68 at 50 years to 1.93 at 85 years.26 In the model, an increased risk of subsequent fractures was also modeled for women who had a prior fracture at the same location.10, 14
Age-specific mortality rates (estimated in 2007–2009) were obtained from the National Institute of Statistics. According to data from a recent meta-analysis,27 hip fractures increased female death probabilities by 4.535 in the first 6 months following the fracture, by 1.755 in the 6 to 12 months period, and by 1.779 in subsequent years. Because the increased mortality following clinical vertebral fractures has been found in many studies to be very similar to those of a hip fracture,28–31 the same impact was assumed after hip and clinical vertebral fractures. Because excess mortality may also be attributable to comorbidities, we conservatively assumed that only 25% of the excess mortality following a hip or vertebral fracture could be directly or indirectly attributable to the fractures themselves.29, 30
Fracture cost and disutility
A healthcare payer perspective including direct medical costs was adopted for all cost estimates, as recommended by a Belgian methodological guideline for pharmacoeconomic evaluations.19 All costs were expressed in the year 2010 using the healthcare product price index when necessary, and discount rates of 3% for costs and of 1.5% for health benefits were assumed for the base-case analysis as recommended in Belgium.19 The direct hospitalization cost of hip fracture, administrated in the first cycle following the fracture, was retrieved from the Belgian national database of hospital bills for the year 2007.32 It included the social security cost and the patient out-of-pocket contribution for nursing and residential fees costs only. Extra costs in the year following the hip fracture was derived from the 2000 study of Autier and colleagues,33 which was based on a prospective controlled study including 159 women. These costs, estimated at €8003 (expressed in 2010€), were equally distributed between the first two cycles following the fracture. Hip fractures are also associated with long-term costs. They were based on the proportion of women being institutionalized following the fracture, ranging from 5% (for women aged 60 years) to 30% (for those aged over 90 years).34 Because women might be institutionalized later in life, regardless of their hip fracture, an adjustment was made to only include long-term costs attributable to the fracture itself.10 The cost of non-hip fractures were quantified relative to hip fracture cost.35 Non-hip fractures were not associated with long-term costs.
Utility values in the general women population as well as relative reductions due to fractures in the year following the fracture and in subsequent years were derived from a systematic review.36 In the case of an occurrence of a second fracture at the same site, the disutility applied to the first fracture event was reduced by 50%.10 This assumption is supported by recent studies showing that the number of fractures is a significant determinant of quality of life.37, 38
The effects of bazedoxifene and raloxifene on fracture risk were derived from the 3-year results of a randomized, double-blind, placebo-controlled and active-controlled study, including healthy postmenopausal women with osteoporosis (55–85 years of age).2 Treatment effects were determined in the overall population and in a subgroup of women at higher fracture risk (femoral neck T-score ≤ −3.0 and/or ≥1 moderate or severe vertebral fracture or multiple mild vertebral fractures) (Table 2). The treatment effect on nonvertebral fractures was assumed in the model for all nonvertebral fractures, including those at the hip and forearm. Subjects were assumed to receive treatment for a maximum of 3 years, as in the phase 3 clinical trial.2 Adherence was not included in the model because no major differences between the two drugs were expected.
Table 2. Relative Risk of Fracture for Bazedoxifene and Raloxifene Versus Placebo2
BZA = bazedoxifene; RLX = raloxifene; CI = confidence interval.
Women with femoral neck T-score ≤ –3.0 and/or ≥1 moderate or severe vertebral fracture or multiple mild vertebral fractures.
Drug costs were based on the official listings of the Belgian Centre for Pharmacotherapeutic Information (September 2011).39 The annual cost of both bazedoxifene and raloxifene was estimated at €406.0 (based on €93.43 for a package of 84 Evista tablets). Additional therapy costs included one physician visit per year (€22.67 per visit) and one bone density measurement at years 1 and 3 (€58.05, including bone densitometry [€35.38] plus an additional physician visit [€22.67]).
Treatment with bazedoxifene and raloxifene were both associated with an increase of 1.9 in the incidence of VTE.2 To account for this in the analysis, VTE was included as a health state in the model during the treatment period. The annual absolute risk of VTE without treatment was estimated at 0.30% in the placebo group of the pivotal trial,2 similar to a large population-based study in the Unites States at the same age range.40 In the model, VTE was assumed to be associated with a 10% utility loss the first year after the event and any utility loss in the second or following years after the event, in agreement with previous health economic publications.5, 8 Mortality rate after VTE was 4.5%41 and the cost of VTE was estimated at €2622, using Belgian estimates of resource utilization based on panel experts.42
The impact of bazedoxifene and raloxifene on breast cancer was not incorporated in the base-case model, as there are no differences between the two treatments in the pivotal study.2 However, in a separate analysis, we considered the raloxifene effects on reducing breast cancer from the Multiple Outcomes of Raloxifene (MORE) trial. In the MORE study, raloxifene reduced the risk of all breast cancer, compared with placebo, by 62% (95% confidence interval: 42%–76%).43 The annual incidence of all breast cancer in Belgian women was obtained from the Belgian Cancer Registry for the year 2009 (between 0.35% and 0.41% for women aged over 60 years) and mortality rate was estimated at 18.6%.44 The direct cost of breast cancer in Belgium (including additional healthcare costs attributable to breast cancer per patient over a 6-year period) was estimated at €14,126.45 Published empirical estimates of health utilities on the effects of breast cancer were included in the analysis.46, 47
Analyses and presentation of results
Incremental cost and incremental effectiveness between bazedoxifene and raloxifene were estimated and represented on a cost-effectiveness plane, where the incremental cost was plotted on the vertical axis and the incremental effectiveness was plotted on the horizontal axis (see Fig. 1 for illustration). The comparator (in this case, raloxifene) is located at the origin. If bazedoxifene is located in the southeast quadrant, it is more effective and less costly and said to be dominant. On the opposite quadrant (northwest), bazedoxifene is more costly and less effective than raloxifene and is therefore dominated. In the other quadrants, bazedoxifene is more effective and more costly than raloxifene (northeast quadrant), or less effective and less costly (southwest quadrant). In order to draw a conclusion about bazedoxifene's cost-effectiveness, the incremental cost-effectiveness ratio (ICER), which is defined as the difference in terms of costs between the interventions divided by their difference in effectiveness and that is represented by the slope of the line between the point and the origin, has to be compared with a cost-effectiveness threshold. This threshold represents the maximum amount that the decision makers are willing to pay per quality-adjusted life-year (QALY) gained. Bazedoxifene would be deemed cost-effective if the ICER falls below the threshold and not cost-effective otherwise.
Probabilistic modeling were undertaken to examine the effect of uncertainty surrounding the model variables. For each model, 200 Monte Carlo simulations were run with 100,000 subjects per simulation.14 Parameter values were randomly selected from their respective distributions for each simulation. Log-normal distributions were assumed, as recommended,48 for relative risk parameters such as fracture risk reduction with therapy and mortality excess. Gamma distribution was used for the incidence of hip fracture and beta distributions were assumed for the effects of fracture on QALYs. Normal distributions, with an SD assumed to be 15% of the mean, were used for the incidence of non-hip fractures, VTE, and breast cancer, the probability of being admitted to nursing home and cost variables given a standard error was not available for these parameters. Using probabilistic modeling, 200 incremental costs and increment effectiveness between the two drugs were then estimated and represented in the cost-effectiveness plane. Cost-effectiveness acceptability curves were also constructed and show the probability that bazedoxifene is cost-effective compared with raloxifene, for a range of decision-maker's willingness to pay.
Sensitivity analyses were conducted to explore the robustness of the results and the degree of impact of model inputs and assumptions on the results. So univariate sensitivity analyses were performed, varying each of the key parameters, including discount rates, fracture disutility, fracture cost, and fracture risk. Analyses were also assessed in women with prevalent vertebral fractures (using a previously described approach14) and at different starting age of treatment. The effect of reducing the cost of raloxifene and of incorporating the raloxifene effects on breast cancer was also investigated.
Overall clinical trial
The cost-effectiveness analysis based on efficacy data from the overall clinical trial indicated that bazedoxifene and raloxifene were equally cost-effective (Fig. 1). Bazedoxifene and raloxifene were each dominant (lower cost for higher effectiveness) in around 40% of the simulations.
Women at higher risk of fractures
When the results were examined based on efficacy data from the subgroup analysis of women at higher risk of fractures, bazedoxifene was dominant (lower cost for higher effectiveness) compared with raloxifene in 84% of the simulations (Fig. 2A). The cost-effectiveness acceptability curve demonstrated that, regardless of the willingness to pay per QALY gained, bazedoxifene was cost-effective compared with raloxifene in approximately 90% of the simulations (Fig. 2B).
Similar results were observed among women aged 70 years with prevalent vertebral fractures (data not shown). Bazedoxifene was dominant (increased effectiveness and decreased cost) in 81% of the simulations and the cost-effectiveness acceptability curve demonstrated that bazedoxifene was cost-effective compared with raloxifene in approximately 90% of the simulations. Age at treatment initiation had also very limited on the results. Among women with BMD T-score ≤ −2.5, bazedoxifene was dominant in 84%, 84%, and 93% of simulations for women aged 60, 70, and 80 years, respectively.
A separate sensitivity analysis evaluated the cost-effectiveness of bazedoxifene compared with raloxifene in which the cost of raloxifene was reduced by 15%, 30%, and 50%. A lower cost of raloxifene decreased the incremental cost between bazedoxifene and raloxifene, and could therefore alter the cost-effectiveness of bazedoxifene. However, even when the cost of raloxifene was reduced by 50%, the probability of bazedoxifene being dominant remained superior to 50% (Fig. 3). At commonly-accepted thresholds for cost-effectiveness (between €20,000 and €80,000 per QALY gained19), bazedoxifene remained cost-effective in more than 80% of the simulations. Additional sensitivity analyses indicated that the effects of increasing or decreasing the fracture risk, fracture disutility, and fracture costs by 25% had no impact on the cost-effectiveness of bazedoxifene compared with raloxifene (Fig. 4).
When incorporating the effect of raloxifene on breast cancer, both treatments were equally effective in terms of QALYs (Fig. 5), meaning that the QALY gain of bazedoxifene resulting from improved antifracture efficacy was compensated by the raloxifene effects on reducing breast cancer. Bazedoxifene is, however, the less costly option in 84% of the simulations, making bazedoxifene more often the dominant option compared to raloxifene (47% versus 15%). At a threshold of €35,000 per QALY gained, bazedoxifene was cost-effective in 68% of the simulations. Even when reducing the cost of raloxifene by 30% or increasing the costs of breast cancer by 50%, bazedoxifene remained, for similar effectiveness, the less costly option in 66% and 83% of the simulations, respectively.
This study investigated the cost-effectiveness of bazedoxifene compared with raloxifene in the treatment of postmenopausal osteoporotic women using treatment effects from a single randomized, double-blind, placebo-controlled and active-controlled study.2 In the overall clinical trial, bazedoxifene and raloxifene have equal clinical effectiveness producing similar cost-effectiveness outcomes. Among women at higher risk of fractures (femoral neck T-score ≤ –3.0 and/or ≥1 moderate or severe vertebral fracture or multiple mild vertebral fractures), bazedoxifene demonstrated improved antifracture efficacy over raloxifene2 that results in a substantial economic value. Bazedoxifene was shown to be dominant (lower cost for higher effectiveness) compared with raloxifene in most of the simulations.
Sensitivity analyses indicated that the results were largely independent of fracture risk, fracture costs, fracture disutility, discount rates, and the age of starting treatment. Findings from this study are therefore generalizable to other populations (at least those with high risk of fractures) and are transferable to other settings (eg, societal perspective) and countries. In addition, a separate sensitivity analysis revealed that bazedoxifene is predicted to remain cost-effective compared with raloxifene among higher-risk women, even if the cost for raloxifene will be reduced as a result of the introduction of a generic version. Our main findings were even robust when rejecting the assumption of a class effect of SERMs on breast cancer.
Previous studies estimated the cost-effectiveness of bazedoxifene for the treatment of osteoporosis women using the FRAX tool.5, 6 As no FRAX-based efficacy data are available for raloxifene, they were restricted to a comparison with no treatment. Depending on country, the base-case results conducted for a 70-year-old woman with a T-score of −2.5 and a prior fragility fracture ranged between “cost-saving” in Sweden and €105,450/QALY in Spain.5 Based on these studies, bazedoxifene can be considered as a cost-effective treatment compared with no treatment for postmenopausal osteoporosis women. However, decision-makers are more interested in comparisons between active drugs in order to determine first-line options. This is especially true because other osteoporotic agents such as raloxifene have been shown to be cost-effective compared with no treatment.8, 9 Our study provides the first pharmacoeconomic evaluation of bazedoxifene compared with an active comparator, suggesting that bazedoxifene is not only cost-effective compared with no treatment, but also a cost-effective option versus raloxifene. This analysis also provides the first cost-effectiveness analysis between active comparators in osteoporosis that used efficacy data from a single trial. Prior cost-effectiveness analyses between active comparators11, 14, 49 used indirect comparisons of efficacy between treatments, which are less robust because of different baseline characteristics of the populations studied and overlapping confidence intervals for the effect of treatment.50
Other osteoporotic treatments are available for the treatment of osteoporosis such as bisphosphonates, denosumab, and strontium ranelate.51 Comparison with these drugs should be carried in the future. However, because no direct comparison between bazedoxifene and these drugs are available, indirect comparisons of efficacy between drugs would be needed, which would be less reliable, as explained in the previous paragraph.
Adherence was not included in this study despite the economic implications of poor adherence to osteoporosis medications and the importance of integrating medication adherence in pharmacoeconomic analyses in osteoporosis.52, 53 However, no major differences between bazedoxifene and raloxifene are expected in terms of medication adherence.
Results of this study suggest that bazedoxifene should be targeted preferentially to women at high risk of fracture. These results are in line with those from the study by Kanis and colleagues,54 which has revealed that bazedoxifene significantly decreased the risk of all clinical fractures and morphometric vertebral fractures in women at or above a FRAX-based fracture probability threshold and that the efficacy increases with increasing fracture probability.
In conclusion, under the assumption of improved antifracture efficacy of bazedoxifene over raloxifene in women with high risk of fractures, bazedoxifene can be considered a cost-effective, and even dominant, pharmacologic option compared with raloxifene in the treatment of postmenopausal women with osteoporosis. Findings from this study suggest that bazedoxifene may be a promising, cost-effective new therapy for the treatment of postmenopausal osteoporosis.
MH has received research grants and/or lecture fees from Amgen, Novartis, Pfizer, Servier, and SMB. JYR has received consulting fees or payments for participating in advisory boards for Servier, Novartis, Negma, Lilly, Wyeth, Amgen, GlaxoSmithKline, Roche, Merckle, Nycomed, NPS, Theramex, and UCB; lecture fees when speaking at the invitation of Merck Sharp and Dohme, Lilly, Rottapharm, IBSA, Genevrier, Novartis, Servier, Roche, GlaxoSmithKline, Teijin, Teva, Ebewee Pharma, Zodiac, Analis, Theramex, Nycomed, and Novo Nordisk; and grant support from Bristol Myers Squibb, Merck Sharp & Dohme, Pfizer, Rottapharm, Teva, Lilly, Novartis, Roche, GlaxoSmithKline, Amgen, and Servier. WBS states that she has no conflicts of interest.
This study was sponsored by Pfizer. The Department of Epidemiology, Public Health and Health Economics, at the University of Liège, received a grant from Pfizer for conducting the study and the development of this manuscript.
Authors' roles: Study design: MH and JYR. Study conduct: MH, WS, and JYR. Data collection: MH and WS. Data analysis: MH and JYR. Data interpretation: MH and JYR. Drafting manuscript: MH. Revising manuscript content: MH, WS, and JYR. Approving final version of manuscript: MH, WS, and JYR. MH takes responsibility for the integrity of the data analysis.