Osteoporosis is a common disease characterized by low bone mass and deterioration of bone tissue, resulting in increased fracture risk.1, 2 The number of osteoporotic fractures in Japan has been increasing as its society ages.3, 4 Osteoporotic fractures are associated with increased morbidity and mortality, and impose a huge financial burden on healthcare systems.5–7 Preventing osteoporotic fractures in the elderly therefore represents an important issue in health economics. To date, the efficacy and cost-effectiveness of osteoporosis treatments, such as alendronate, have been studied extensively.8–11 However, it is less clear whether postmenopausal women with osteopenia should be treated with pharmaceuticals. One study from Europe showed that the cost-effectiveness of alendronate in osteopenic women without a history of fracture varies depending on the setting of each country, discount rate, and willingness to pay thresholds.12 Another study from the United States showed that alendronate is not cost-effective for postmenopausal women with osteopenia who do not have additional risk factors of fracture.13
Recently, the revised version of Japanese guidelines for the treatment of osteoporosis has been released, which recommend therapeutic intervention in postmenopausal women with osteopenia without a history of fracture: (1) if they have a family history of hip fracture, or (2) if the 10-year probability for osteoporosis-related fracture, as estimated by the WHO fracture risk assessment tool (FRAX), is more than 15%.14 However, no studies to date have examined the cost-effectiveness of treatment to prevent fractures in osteopenic postmenopausal women from the Japanese perspective. Evidence from the previous studies may not be applicable to Japanese population given the differences in epidemiological characteristics and the healthcare system. Although the prevalence and incidence of vertebral fracture in Japan are higher than those in the United States and Europe,15, 16 the incidence rate of hip fracture, which may cause a greater burden than other osteoporotic fractures, is lower in Japan compared to Western countries.3, 17 Here we estimated the cost-effectiveness of alendronate therapy to prevent fractures in postmenopausal Japanese women who have low bone mineral density (BMD) but do not have osteoporosis or a history of fracture. Because the FRAX algorithm is not open to the public, we performed a model-based, cost-effectiveness analysis that considered age, BMD, and the presence of clinical risk factors by using published sources.
Materials and Methods
The modeled population was osteopenic postmenopausal Japanese women who do not have a history of fracture. The base case was 70-year-old women with a BMD of 70% of young adult mean (YAM) having a family history of hip fracture. The current Japanese guidelines define osteopenia as a femoral neck BMD between 70% and 80% of the YAM. In this model, the value of YAM was set to 0.787 g/cm2 (SD = 0.109 g/cm2), which was derived from a survey in Japan.18 BMD values between 70% and 80% of YAM (0.551 and 0.630 g/cm2) in our model are equivalent to T-scores of −2.56 and −1.90, respectively, when using the reference mean of 0.858 g/cm2 (SD = 0.120 g/cm2) for non-Hispanic white women aged 20 to 29 years, derived from the National Health and Nutrition Examination Survey (NHANES) III data.18, 19
The model-based, cost-effectiveness analysis was conducted to evaluate the cost-effectiveness of 5 years of preventive alendronate therapy compared with no preventive therapy in postmenopausal women with osteopenia in Japan. A patient-level state transition model was developed to predict long-term costs and quality adjusted life years (QALYs) associated with preventive alendronate therapy and no preventive therapy (Fig. 1). The model consists of six health states: “no fracture,” “postvertebral fracture,” “post-hip fracture,” “postvertebral and post-hip fracture,” “bedridden,” and “death.” Fractures other than hip and clinical vertebral fractures were not considered in the analysis. In the model simulation, subjects started with the state of “no fracture” and faced different risks of fracture depending on their age, femoral neck BMD, presence of BMD-independent clinical risk factors (ie, a family history of hip fracture, high alcohol intake, and current smoking), and the treatment received. The cycle length of the model was set to 1 year. During each cycle of the model, subjects experienced one of the following clinical events: “no event,” “vertebral fracture,” “hip fracture,” or “both vertebral and hip fracture.” Once subjects had a fragility fracture, they moved to the health states of post-fracture and had an additional risk for subsequent fractures. In preventive therapy arm, all subjects started to receive 5-year alendronate therapy to prevent fracture. In no preventive therapy arm, subjects did not receive treatment to prevent fracture, but if a fragility fracture occurred they started to receive alendronate during next 5 years to prevent subsequent fractures. We assumed that a certain proportion of subjects having a hip fracture moved to the health state of “bedridden.” The model was developed and analyzed using TreeAge Pro 2011 (TreeAge Software, Williamstown, MA, USA).
A transition probability (p) of an event occurring over a time interval (t) was calculated using an incidence rate (r) according to the following formula: p = 1–exp(−rt).20 We developed equations for age and femoral neck BMD-specific fracture rate by using a series of methods by De Laet and colleagues21 and epidemiological data of postmenopausal Japanese women with osteopenia or osteoporosis.4, 16, 18 (see Supplemental Appendix). We first developed equations for age-dependent fracture rate for vertebral fracture (with or without a previous fracture) and hip fracture by using data on postmenopausal Japanese women and curve-fitting techniques.4, 16 An exponential curve that showed the highest R2 value was used for fitting the model. We then estimated age- and BMD-specific fracture rates by combining the equations for age-dependent fracture rate, distributions of femoral neck BMD by age group, and relative risks (RRs) of fracture per 1 SD reduction in BMD (Table 1).
Table 1. Equations for Estimating Probabilities of Osteoporotic Fracture and Death Related to Hip Fracture
BMD, bone mineral density; Z, Z-score.
Equation for age and BMD dependent incidence rate of fracture
Incident rates of subsequent fractures for those with a previous fracture were calculated by multiplying the age- and BMD-specific fracture rate by the RR of subsequent fracture 22 (Table 2). Fracture rates for those with additional clinical risk factors were calculated in the same manner by multiplying by the RR of each clinical risk factor23–25 (Table 2). In this study, we modeled clinical vertebral fractures and did not consider nonsymptomatic morphometric vertebral fractures. The proportion of clinical fractures among all vertebral fractures was assumed to be 30% for the base case analyses (Table 2).
Table 2. Parameter Inputs for Transition Probabilities
Age-dependent mortality rate was obtained from the life table reported by the Ministry of Health, Labour, and Welfare in Japan.26 The odds ratio for death 10 years after hip fracture was formulated by using curve-fitting derived from a survey in Japan27 (Table 1). The mortality rate for subjects who experienced a hip fracture was calculated by multiplying the age-dependent mortality by the odds ratio of death following hip fracture. The probability of becoming bedridden after hip fracture was derived from a published source in Japan.28
We considered 5 years of alendronate treatment at a dosage of 35 mg per week for the base case scenario. Because no studies have specifically examined the effect of alendronate to prevent fracture in postmenopausal Japanese women with osteopenia, we used data derived from a recent meta-analysis of three randomized controlled trials of alendronate therapy in postmenopausal women with osteopenia or osteoporosis conducted in countries other than Japan.8 We modeled residual effects of alendronate, assuming a linear decline in efficacy over 5 years, after 5 years of treatment.13, 29 In this model, full compliance with alendronate treatment was assumed, and loss of efficacy due to partial adherence was considered in a sensitivity analysis using the following formula: efficacy in subjects with partial adherence = (1–[1–RR of fracture with alendronate] × [1–percent reduction in efficacy]).
We considered only direct medical costs from the perspective of the Japanese healthcare system. Table 3 summarizes input values for cost parameters. All costs were calculated in Japanese yen and converted to U.S. dollars with the currency exchange rate of $1 = JPY80 as in July 2012. The annual drug cost of alendronate was calculated by using a Japanese price list for drugs.30 Medical costs for daily practice, including consulting, clinical testing, and radiography, were estimated on the basis of Japanese tariffs,31 assuming standard clinical practice.28 Medical costs due to a fracture event were obtained from published sources.7, 32 Because no studies provided data on the costs of nursing care for a bedridden patient, we used data for patients provided with nursing care level 5 under the Nursing Care Insurance Scheme in Japan (those who are unable to perform daily life activities without assistance).33
Input values for utility parameters are shown in Table 3. The utility values, based on the EuroQOL-5D (EQ-5D) for event-free women by age group, were derived from Japanese population data.34 The utilities of patients who experienced fracture events were calculated by multiplying the event-free utility value by the percent reduction in utility associated with various fracture events.35 Given the lack of data on utility values for bedridden patients, we used data for patients provided with nursing care level 5.36
Base case analysis
For the base case analyses, a first-order Monte Carlo simulation (individual simulation) using 1000 women aged 70 years with a femoral neck BMD of 70% of YAM was performed with 1000 iterations to obtain the point estimate of lifetime expected costs and QALYs associated with preventive alendronate therapy compared with no preventive therapy. An annual discount rate of 3% for both costs and health benefits was applied. The incremental cost-effectiveness ratio (ICER) was estimated by the following formula: ICER = (Costpreventive alendronate therapy − Costno preventive therapy)/(QALYpreventive alendronate therapy − QALYno preventive therapy).
Deterministic sensitivity analyses were performed to assess the influence of varying key parameters on the base case results. Assessed parameters and ranges are shown in Tables 1 to 3. The plausible ranges for each parameter were determined based on reported values, such as 95% confidence intervals in published sources, or expert opinion. To estimate the influence of low compliance on cost-effectiveness, we modeled a 20% reduction in efficacy associated with partial adherence in the sensitivity analysis.37 We also performed scenario sensitivity analyses with different combinations of the starting age (65, 70, and 75 years), femoral neck BMD (70%, 75%, and 80% of YAM), and the number of clinical risk factors (one, two, and three). Based on the results of scenario sensitivity analyses, the relationship between the ICER and the 10-year probability of an osteoporotic fracture was analyzed using curve fitting. Probabilistic sensitivity analyses were performed by using 1000 iterations of a second-order Monte Carlo simulation to examine the influence of parameter uncertainty on the cost-effectiveness. Probabilistic distributions of the parameters used in probabilistic sensitivity analyses are shown in Table 2. We applied a relevant distribution for each variable in the model if data on variability was available. In each of the 1000 simulations, the value for each model input was randomly selected from its distribution. Based on the results of the second-order Monte Carlo simulation, we constructed cost-effectiveness acceptability curves and estimated the proportion of iterations in which alendronate treatment would be preferred in terms of cost-effectiveness, assuming a willingness to pay threshold of $50,000 per QALY gained.
Base case analysis
The results of the cost-effectiveness of alendronate therapy for osteopenic 70-year-old women with a BMD of 70% of YAM are shown in Table 4. For women with no clinical risk factors who did not receive treatment for primary fracture prevention, the 10-year probability of hip fracture and the 10-year probability of vertebral fracture were estimated to be 3.27% and 8.41%, respectively. Five years of alendronate treatment followed by 5 years of residual effects resulted in a 28.8% risk reduction in hip fracture and a 34.3% risk reduction in vertebral fracture. Compared to no preventive therapy, preventive alendronate therapy incurs average additional lifetime costs of $7,977 per person and confers an additional 0.035 QALYs, which resulted in an ICER of $227,905 per QALY gained. For women with one clinical risk factor who did not receive preventive treatment, the 10-year probability of hip fracture and vertebral fracture ranged from 4.55% to 7.64% and from 11.61% to 15.30%, respectively. In the preventive alendronate treatment group, the 10-year probability of hip fracture and vertebral fracture ranged from 3.27% to 5.49% and from 7.63% to 10.04%, respectively. Alendronate treatment resulted in a 28.1% to 28.2% risk reduction in hip fracture and a 34.1% to 34.4% risk reduction in vertebral fracture. The ICERs for alendronate treatment in osteopenic women with a family history of hip fracture (base case), those with high alcohol intake, and those with current smoking were estimated to be $92,937, $126,251, and $129,067 per QALY gained, respectively.
Table 4. Cost-Effectiveness of Alendronate Therapy for 70-Year-Old Women With a BMD 70% of YAM
Ten year probability of fracture (%)
Incremental Cost (US$)
ICER (US$ per QALY)
BMD, bone mineral density; YAM, young adult mean; QALY, quality-adjusted life year; ICER, incremental cost-effectiveness ratio.
Without risk factor
No preventive therapy
Preventive alendronate therapy
With family history of hip fracture (base case)
No preventive therapy
Preventive alendronate therapy
With alcohol intake (>2 unit per day)
No preventive therapy
Preventive alendronate therapy
With current smoking
No preventive therapy
Preventive alendronate therapy
The results of deterministic sensitivity analyses for the base case were summarized by using a tornado diagram as shown in Fig. 2. The RR of hip fracture associated with alendronate therapy had a relatively large impact on the base case result, but the ICERs remained higher than $50,000 per QALY over the full range of model parameters. Similarly, the ICER was above $50,000 per QALY gained in all scenarios for 70-year-old women with a BMD of 70% of YAM having current smoking or alcohol intake (data not shown).
The results of scenario sensitivity analyses with different combinations of age, BMD, and number of clinical risk factors are shown in Fig. 3. Higher age, lower BMD, and a larger number of clinical risk factors were associated with a decrease in the ICER. For women with one clinical risk factor, the ICER of preventive alendronate therapy was above $50,000 per QALY in all scenarios. For women with two clinical risk factors, the ICER was below the acceptable level in the following scenarios: (1) 75-year-old women with a BMD of 70% of YAM having a family history of hip fracture and high alcohol intake; and (2) 70- and 75-year-old women with a BMD of 70% of YAM having a family history of hip fracture and smoking. For women with all three risk factors, the ICER was below the acceptable level in the following scenarios: (1) all women with a BMD of 70% of YAM; and (2) 70- and 75-year-old women with a BMD of 75% of YAM.
The relationship between the ICER and 10-year probability of an osteoporotic fracture was analyzed by using exponential curve fitting. As shown in Fig. 4, the ICER for preventive alendronate therapy was less than $50,000 and $100,000 per QALY gained when the 10-year fracture probability was more than 26.2% and 19.1%, respectively.
Probabilistic sensitivity analyses were performed for the base case of 70-year-old women with a BMD of 70% of YAM and a family history of hip fracture. Cost-effectiveness acceptability curves constructed on the basis of probabilistic sensitivity analyses are shown in Fig. 5. Using a willingness to pay a threshold of $50,000 per QALY gained, the probability that 5 years of alendronate therapy became cost-effective compared to no preventive therapy was estimated to be 0.2% to 2.6%, 13.1% to 56.1%, and 99.1% for those with one, two, and three clinical risk factors, respectively.
In this economic evaluation, we estimated the cost-effectiveness of 5 years of preventive alendronate therapy relative to no preventive therapy for osteopenic postmenopausal Japanese women who do not have a history of clinical fracture. Consistent with the study by Schousboe and colleagues,13 we found that preventive alendronate therapy was unlikely to be cost-effective for osteopenic women who had no clinical risk factors, assuming a societal willingness to pay of $50,000 per QALY. It is, however, noteworthy that the ICER of alendronate therapy for osteopenic women who have no previous fracture and no additional risk factors in our analysis appears to be higher than that of previous studies. While the ICER in 70-year-old Japanese women who have no risk factors and a BMD of 70% of YAM (equivalent to a T-score of −2.56) was $227,905 per QALY gained, the ICER in women with a T-score of −2.4 ranged from $74,200 to $86,465 per QALY in the U.S. study.13 In the European study, the ICER in 69-year-old women with a T-score of −2.5 was lower than €80,000 ($60,000, if €1 = $0.75) per QALY.12 These differences are consistent with the results from another study from the United States showing that osteoporosis treatment in white populations is likely to be more cost-effective than in Asian populations.38 There are several possible explanations for this gap, one of which is the difference in epidemiological characteristics such as the lower incidence rates of hip fracture in Japanese populations compared to their white counterparts.3, 17
In this study, we also examined the cost-effectiveness of preventive alendronate therapy with different combinations of age, BMD, and clinical risk factors. We found that the cost-effectiveness of preventive alendronate therapy is sensitive to variations in age, BMD, and number of clinical risk factors. Probabilistic sensitivity analyses showed that the probability of being cost-effective was 0.2% to 2.6%, 13.1% to 56.1%, and 99.1% in 70-year-old women with a BMD of 70% of YAM who have one, two, and three clinical risk factors, respectively. Our results suggest that the decision to treat osteopenic women with alendronate should be made based on the age, BMD, and number of clinical risk factors in terms of cost-effectiveness.
Although the FRAX is considered to be valid and reliable and the Japanese version of the FRAX has also been developed,39, 40 the algorithm is not open to the public and is thus unavailable for economic evaluations. In the present study, we developed risk equations for age- and BMD-specific incidence rate of fracture using epidemiological data from the Japanese population and combined them with the state transition model. The validity of our simulation model was verified by comparison of the predicted 10-year osteoporotic fracture probabilities in our model and those derived from the FRAX.39 For 70-year-old women with different combinations of T-scores (−1.5, −2.0, and −2.5) and number of clinical risk factors, the estimated 10-year probabilities of hip fracture in our model were almost identical to those of the FRAX (Supplemental Fig. 1). The probabilities of major fracture in our model also appears to be consistent with those of the FRAX, as our model considered only hip fracture and clinical vertebral fractures whereas the definition of major fracture in the FRAX includes hip, clinical vertebral fracture, humerus, and wrist fractures (Supplemental Fig. 2). These findings support the validity of our model in this economic analysis.
Additional analyses on the relationship between an ICER and the predicted 10-year fracture probability indicated that an ICER was below $50,000 or $100,000 per QALY if the 10-year fracture probability was above 26.2% or 19.1%, respectively. Although there is a discrepancy in the definition of major fracture between our model and the FRAX, our findings may have important implications for considering the cost-effective treatment for osteopenic postmenopausal women in Japan. The latest Japanese guidelines recommend drug therapy for osteopenic postmenopausal women whose 10-year fracture probability, based on the FRAX, is more than 15%.14 Our results suggest that in terms of cost-effectiveness, the current treatment threshold is not acceptable even if Japanese society would be willing to pay $100,000 per additional QALY.
The main limitation of our analysis is that we did not consider other osteoporotic fractures such as humerus, wrist, and radiographic vertebral fracture because of a lack of data. Our model may underestimate the true cost-effectiveness of alendronate treatment in osteopenic menopausal women because it may have underestimated the total risk of osteoporotic fractures. Although the impact of other osteoporotic fractures on health outcomes seems to be relatively small compared to that of hip fracture and clinical vertebral fracture,35 the results of our analysis should be interpreted with caution. Another limitation is that we estimated the efficacy of alendronate using the meta-analysis of randomized clinical trials that were conducted in countries other than Japan. Given that the sensitivity analysis showed that the uncertainty about the efficacy of alendronate on the RR for hip fracture had a relatively large impact on the ICER of preventive alendronate therapy, the efficacy of alendronate in postmenopausal Japanese women with osteopenia should be determined to confirm the validity of our results. Finally, it is important to note that the cost-effectiveness of alendronate in osteopenic women varies depending on the willingness to pay thresholds for each additional QALY gained. Although the willingness to pay threshold of $50,000 per QALY and £20,000 to £30,000 per QALY has commonly been used as the acceptable level in the United States and the United Kingdom,41, 42 respectively, it should vary from country to country. Although still controversial, a study by Ohkusa and Sugawara43 has proposed a willingness to pay threshold of JPY6,350,000 to JPY6,700,000 ($79,375 to $83,750) per QALY gained, and the cost-effective intervention for osteopenia may vary depending on the acceptability level of the ICER.
In conclusion, the cost-effectiveness of preventive alendronate therapy for osteopenic women is sensitive to age, BMD, and the number of clinical risk factors. Our analysis suggests that the treatment indication for osteopenic postmenopausal women without a history of fracture in the current Japanese guidelines is unlikely to be cost-effective if society is willing to pay $50,000 per additional QALY. In terms of cost-effectiveness, preventive alendronate therapy should be considered for a more selected population on the basis of age, BMD, and number of clinical risk factors.
KM has received grant support from the Pfizer Health Research Foundation. HK has received honoraria from Kyowa Hakko Kirin, Chugai Pharmaceutical, and Bayer Yakuhin. MF has received honoraria and grant support from Kyowa Hakko Kirin, Chugai Pharmaceutical, and Bayer Yakuhin. All of the other authors state that they have no conflicts of interest.
This study was supported by a research grant from the Pfizer Health Research Foundation.
Authors' roles. Study design: KM, HK, SN, MF, and HT. Data collection: KM, HK, SN, SY, HI, and T.Toujo. Data analysis: KM. Data interpretation: KM, HK, SN, SY, T.Takiguchi, T.Toujo, MF, and HT. Drafting manuscript: KM and HK. Revising manuscript content: SN, SY, T.Takiguchi, HT, T.Toujo, MF, and HT. Approving final version of manuscript: KM, HK, SN, SY, T.Takiguchi, HI, T.Toujo, MF, and HT. KM takes responsibility for the integrity of the data analysis.