Cost-effectiveness of denosumab for high-risk postmenopausal women with osteoporosis in Thailand

Abstract

Aims: This study assessed the cost-effectiveness of denosumab for treating postmenopausal women with osteoporosis (PMO) at high risk of fracture in Thailand.

Materials and methods: A published Markov cohort cost-effectiveness model was populated with country-specific data as available and other published data as needed. The model used a societal perspective, lifetime horizon, efficacy data from network meta-analysis of trials, and included costs for direct medical and non-medical care, informal care, and osteoporosis treatments to compare denosumab to no pharmacologic treatment (calcium and vitamin D supplements only) and to oral weekly alendronate. The base case (high-risk population) included postmenopausal women with femoral neck T-score ≤−2.5, mean age 65 years at entry, and history of vertebral fracture.

Results: High-risk women with osteoporosis using denosumab had the greatest number of life years and quality-adjusted life-years (QALYs) with higher reductions in hip and vertebral fracture incidence compared with patients with no pharmacologic treatment. The incremental cost-effectiveness ratio (ICER) was 119,575 Thai Baht (THB) per QALY for denosumab versus no pharmacologic treatment and 199,186 THB per QALY for denosumab versus alendronate. Among Thai postmenopausal women with high-risk of fractures, denosumab was cost-effective compared with no pharmacologic treatment at a willingness-to-pay threshold of 160,000 THB per QALY. One-way sensitivity analysis showed models were most sensitive to changes in denosumab pricing.

Limitations: Data from other countries used when country-specific data were unavailable may not accurately reflect the true experience in Thailand. The model focused explicitly on hip, vertebral, and wrist fractures, and therefore provides a conservative estimate of the overall potential impact of osteoporosis-related fracture. The fracture risk was not adjusted to reflect potential changes in risk after denosumab treatment discontinuation.

Conclusions: In Thailand, denosumab offers a cost-effective osteoporosis treatment option versus no pharmacologic treatment in postmenopausal women at high risk of fracture.

Keywords:

• denosumab
• osteoporosis
• fracture
• cost-effectiveness
• Thailand

JEL classification codes:

• I10
• I18

Introduction

Postmenopausal osteoporosis is a common disorder associated with significant morbidity and mortality and reduced quality of life^1,2. In Thailand, the age-adjusted prevalence of osteoporosis in postmenopausal women was 19.8% and 13.6% for lumbar spine and femoral neck, respectively, ranging from 5% with aged less than 50 years to more than 50% among women over age 70 years³. As the aged population (60 years and older) in Thailand grows from 14% of the total population in 2015 to nearly 30% by 2050, the number osteoporosis-related fractures, fracture-related healthcare costs, and post-fracture mortality will also rise^4–6. The incidence of hip fracture is projected to increase nearly 3-fold between 2018 and 2050, as the number of fractures grows from 30,560 to 86,765 among Thai postmenopausal women (age 50 and older), while, the incidence of vertebral fracture in women ≥50 years of age in a suburb of Bangkok of 32.1/1,000 person-years⁷.

Anti-osteoporotic agents are well established for the treatment of postmenopausal women at a high risk of fractures, and have been shown to reduce the risk of vertebral, nonvertebral and hip fractures^8,9. Besides the clinical profile of a therapy, it is also important to consider consequences of treatment in terms of costs and health effects both in short- and long-term perspectives. The clinical outcome and cost-effectiveness of anti-osteoporotic drugs have been evaluated and proved using health economic models integrating epidemiological, clinical and economic data from both the payer and societal perspectives in several countries^10–17. However, there is a dearth of data to evaluate the cost-effectiveness of osteoporosis treatment in developing countries. This study was undertaken to assess the cost-effectiveness of denosumab compared with calcium and vitamin D supplements only (i.e. no pharmacologic therapy) or alendronate for treating postmenopausal women with osteoporosis at high risk of fractures in Thailand. In clinical trials, over three years, the RANK ligand inhibitor denosumab was associated with a 61–71% relative risk reduction in the incidence of new vertebral fractures and a 20% relative risk reduction for new nonvertebral fractures¹⁸. Denosumab is one of the newest postmenopausal osteoporosis (PMO) treatment agents approved for use in Thailand, while alendronate, an oral bisphosphonate, is the most commonly used osteoporosis treatment.

Methods

Target population

The target population used in the base-case analysis for this study was postmenopausal women with osteoporosis (PMO) who were at high risk of fracture as defined by a mean age of 65 years, a T-score ≤−2.5 at the femoral neck, and a history of vertebral fracture. A separate model scenario was run to evaluate cost-effectiveness in the overall PMO population. In this scenario, the T-score requirement remained the same as in the base-case, but the average starting age was lowered to age 60 years. The prevalence of vertebral fracture was also lowered from 100% to 23.6% based on local epidemiological data for the general Thai population of women age 60 years and older⁷.

The model

In this study, a previously published Markov cohort cost-effectiveness model¹³ was populated with data specific to Thailand. This model was selected for the current study since it employed a societal perspective with a lifetime horizon (i.e. patients were followed from age at treatment initiation to death or a maximum age of 100 years), included costs for direct medical and non-medical care as well as the cost of informal care, and results derived previously from that model will provide useful context for results from the current study. A six-month cycle length was used in alignment with the recommended denosumab dosing frequency, and for each cycle women transitioned into one of seven mutually exclusive health states (Figure 1). All women entered the model in the well health state, and in each cycle could remain well, sustain a fracture, or enter the dead state. Patients who entered the dead state remained there for the rest of the simulation. Women who sustained a fracture during any cycle would enter either the hip fracture, vertebral fracture, or wrist fracture health state. After one year in any of the fracture health states, the model allowed a woman to either sustain a new fracture or move to either the post-hip or post-vertebral fracture health states or to the dead health state. Women who were in the post-hip fracture health state were only at risk of sustaining new hip fractures or dying whereas women who were in the post-vertebral fracture health state could sustain either new vertebral fractures or hip fractures or die. This approach also means that women who sustained either a hip or vertebral fracture (the two most serious fracture types) were not allowed to have future wrist fractures.

Figure 1. Structure of Markov cohort model.

The model compared three osteoporosis treatment options: (1) denosumab (subcutaneous injection, once every six months); (2) no pharmacologic treatment (daily intake of calcium or calcium and vitamin D supplements only); (3) alendronate (oral, once weekly), which is the most commonly prescribed pharmacologic treatment for women with postmenopausal osteoporosis in Thailand. Treatments were intended to be provided for a duration of five years. The “no pharmacologic treatment” comparator provides insights into cost-effectiveness for patients who cannot take alendronate because of contraindications, or adherence or tolerance issues.

The incremental cost effectiveness ratio (ICER) expressed as cost per quality adjusted life year (QALY) gained is the primary outcome from this cost-effectiveness analysis. The number of patients needed to treat (NNT), calculated as the reciprocal of the absolute reduction in fractures, to avoid one hip fracture in 10 years was also determined for both denosumab and alendronate. The ICER threshold for the analysis was obtained from the Thailand National List of Essential Medicines (NLEM) guidance for health economic evaluation, and set at 160,000 Thai Baht (THB) per QALY¹⁹. For reference, on 31 December 2018, one THB was valued at 0.03 United States dollars²⁰ and this threshold is equivalent to ∼5,300 US dollars/QALY.

Model parameters

A literature review, the scope of which was defined by the Population, Intervention, Comparator, Outcome and Study design (PICOS) framework (Supplemental Table 1), was conducted to identify relevant Thai data from publications indexed in PubMed from the inception of the database through July 2018. These published data were supplemented through discussion with local experts in the clinical management of osteoporosis.

Fracture incidence and mortality

Age-specific estimates of fracture incidence in Thailand, derived from the general population were used in the model and are provided in Table 1. In the model, the risk of sustaining a fracture depends specifically on three elements: (i) the risk for an individual in the general population of incurring a fracture, (ii) the increased fracture risk associated with osteoporosis (the relative risk), and (iii) risk reduction, if any, attributed to a treatment. The fracture risk in the general population was adjusted upwards to adjust for bone mineral density (BMD) and vertebral fracture history in the osteoporotic population, as described in Jönsson et al.¹³. The older average age at model entry in the high-risk population compared with the general population (65 years versus 60 years) implies a higher incidence of hip fracture (16 per 10,000 women versus 8 per 10,000 women), although the incidence of vertebral and wrist fracture were assumed to be the same in both populations. Both hip and vertebral fractures have been shown to increase mortality risk above the background risk in the general population in Thailand²¹. Due to lack of local data, a value of zero was used for all input parameters associated with non-osteoporotic fracture (e.g. risk of event, event-related direct medical and non-medical costs, utility values). With this approach, the “other osteoporotic fractures” arm is still included in the model structure but does not contribute to the model results. For the model, an 18% excess mortality risk was assumed to last for eight years following a hip or vertebral fracture²¹.

Table 1. Fracture incidence per patient-year in Thailand.

Age	Fracture type
	Hip	Vertebral	Wrist
50	0.0003	0.0274	0.0018
55	0.0003	0.0274	0.0032
60	0.0008	0.0373	0.0033
65	0.0015	0.0373	0.0034
70	0.0039	0.0461	0.0036
75	0.0079	0.0461	0.0033
80	0.0124	0.0461	0.0038
>85	0.0141	0.0461	0.0042

Sources: Jitapunkul et al., Wongtriratanachai et al. and Health Intervention Technology Assessment Program (HITAP) Ministry of Public Health [7,44,46].

Costs

All costs were evaluated from a societal perspective and quantified in THB adjusted to 2018 values using the Thai Consumer Price Index (CPI)²². Direct fracture-related costs included all costs related to managing the fracture in the inpatient, outpatient and long-term care settings (Table 2). For analytic purposes, total direct fracture-related costs were split into inpatient (37%) and outpatient (63%) based on results from a prior study²³, although this classification did not impact the total costs. The annual cost of long-term care reported by Khongboon et al.²⁴ was converted into a daily cost of THB 484. Hip and vertebral fractures costs were applied in full for the first year after the fracture, and the cost of four physician visits and one x-ray were assumed for fracture patients in each subsequent year. Wrist fractures were assumed to have an impact only on costs during the first year after the event. Each fracture was assumed to require four physician visits during the first year, regardless of fracture type or patient age. A standard direct non-medical cost (2,929 THB per patient) was assumed for all fracture types with this amount representing the cost of travel and meals associated with the fracture-related physician visits²⁵. The annual per patient cost of informal care was an average of costs in urban and rural areas and assumed to be 72,150 THB, regardless of patient age or fracture type²⁴. The annual cost of denosumab was estimated to be 25,550 THB, and annual alendronate costs, based on weekly oral dosing, were assumed to be 4,570 THB (based on median price per unit and dosing per label²⁶) For patients who sustained a fracture, fracture-specific reductions in utility were applied. Due to the absence of local Thai data, the utility estimates used in this analysis were taken from foreign publications (Table 3)^27,28. Although gastrointestinal adverse events are associated with the use of oral bisphosphonates, such as alendronate, costs associated with treating these events were not included in the base-case scenario. A discount rate of 3% per annum was applied.

Table 2. Summary of costs.

Direct fracture-related medical costs
Type of fracture	Cost (THB)	Source
First year of fracture
Hip	54,245	[51]
Vertebral	38,305	[51]
Wrist	28,885	[52]
Second and subsequent years
Hip	1,756	Calculated based on expert’s opinion
Vertebral	1,756	Calculated based on expert’s opinion
Cost item	Unit cost (THB)	Source
DXA scan	2000	[52]
Physician visit	314	[45,52]
IV injection	32	[45,53]
Nurse visit	0	[43,44]
Direct non-medical	2,929	[25]
Informal care	72,150	[24]
Denosumab, annual cost	25,550	[54]
Alendronate, annual cost	4,570	[54]

Abbreviations. DXA, dual energy x-ray absorptiometry; IV, intravenous; THB, Thai Baht. Costs expressed in 2018 values.

Table 3. Summary of utility estimates.

Age	Utility	Source
50	0.94	[28]
60	0.91	[28]
70	0.91	[28]
80	0.88	[28]
Fracture type/period	Utility estimate	Source
First year after fracture
Hip fracture	0.797	[29]
Clinical vertebral fracture	0.720	[29]
Wrist fracture	0.956	[27]
Second and following years after fracture^a
Hip fracture	0.899	[29]
Clinical vertebral fracture	0.931	[29]

^aWrist fractures were assumed to only have an impact on quality of life during the first year.

Treatment efficacy

Estimates for the treatment efficacy parameters (Table 4) were obtained from the random effects network meta-analysis of data from randomized controlled trials conducted by Freemantle et al.²⁹ which provides robust data by pharmacologic agent and fracture site. Treatment persistence (i.e. continuing treatment according to medical recommendation/prescribing) varies in real-world use and poor persistence can reduce the clinical benefit that a patient derives from a prescribed osteoporosis medication. Persistent use of osteoporosis medications has been associated with greater reduction in fracture rates, lower healthcare utilization, and lower total healthcare costs^7,9,30,31. Therefore, treatment-specific persistence was modelled in each six-month cycle for the duration of the treatment course (Table 5). The assumption was made that only persistent patients incur drug costs and receive the full fracture reduction effect of treatment. Since treatment benefit does not stop immediately following discontinuation, it is assumed that the fracture reduction effect declines linearly to 0 over the course of 2 years after treatment ends. For denosumab, per-cycle persistence was implemented as a step function, to reflect the semi-annual dosing schedule, where all patients remain persistent for six months after an injection. Since oral alendronate is delivered weekly, it was assumed that discontinuation could occur at any point in each half-year cycle. As such, persistence was assumed to be derived from a continuous function, with average persistence taken at the midpoint of each six-month cycle. Since persistence data were not available for every six-month period for either treatment, where required, persistence was interpolated by assuming a constant drop-out rate between observed time points, and extrapolated by assuming a drop-out rate equivalent to the last observed interval.

Table 4. Treatment efficacy: relative risk of fracture (treatment versus placebo).

	Hip fractures RR (95% CI)	Vertebral fractures RR (95% CI)	Wrist fractures RR (95% CI)
Alendronate	0.65 (0.41–1.03)	0.56 (0.46–0.69)	0.81 (0.37–1.80)
Denosumab	0.61 (0.37–0.98)	0.33 (0.26–0.41)	0.84 (0.64–1.11)

Source: random effects meta-analysis [30].

Table 5. Treatment persistence estimates.

Treatment	Time (in months)						Source
	6	12	18	24	36	60
Denosumab	1.00	0.78 [0.73; 0.81]	0.68 [0.65; 0.70]	0.57 [0.52; 0.63]	NA	NA	[55]
Alendronate	0.53 [0.52; 0.53]	0.41 [0.41; 0.42]	NA	NA	0.25 [0.24; 0.25]	0.17 [0.17; 0.17]	[54]

Abbreviation. NA, not available.

Analysis

In addition to producing results for the base-case (high-risk population), results were modelled separately for the general population of postmenopausal women with osteoporosis (overall osteoporosis population). One-way sensitivity analysis was conducted to assess the effect of varying drug and fracture costs, excess mortality risk, and treatment duration by up to 25% from the base-case parameters. Probabilistic sensitivity analysis as described in the Model section above was conducted to assess cost-effectiveness over a range of ‘willingness-to-pay’ values.

Results

In the base-case (i.e. high-risk population) analysis, despite having the highest total cost, denosumab produced the lowest fracture-related cost of all modelled treatment options (Table 6 and Table 7). This is due to cost savings from the lower incidence of hip and vertebral fractures compared with both no pharmacologic treatment and alendronate. In addition, women treated with denosumab were estimated to have a greater number of life years and QALYs compared to those who received no pharmacologic treatment (0.415 and 0.362) and to those treated with alendronate (0.248 and 0.214) (Table 8). The ICER for denosumab versus no pharmacologic treatment was 119,575 THB per QALY (3,587 USD per QALY) and the ICER for denosumab versus alendronate was 199,186 THB per QALY (5,976 USD per QALY). In the model scenario for the total PMO population, the corresponding ICERs were 153,879 THB per QALY (4,616 USD per QALY) (versus no pharmacologic treatment) and 243,983 THB per QALY (7,319 USD per QALY) (versus alendronate).

Table 6. Costs and fracture results for base-case model^a.

	Denosumab	No pharmacologic treatment	Alendronate	Denosumab versus no pharmacologic treatment	Denosumab versus alendronate
Direct fracture-related medical cost (THB)	93,039	104,894	99,430	−11,856	−6,391
Hospitalization cost	24,693	28,666	26,953	−3,973	−2,260
Long- term care cost	26,301	27,419	26,585	−1,118	−284
Outpatient cost	42,045	48,810	45,892	−6,765	−3,848
Direct fracture-related non- medical and informal care costs (THB)	55,233	64,531	60,502	−9,298	−5,269
Intervention cost (THB)	64,424	0	10,171	64,424	54,252
Drug cost	59,187	0	6,810	59,187	52,378
Treatment management	5,236	0	3,362	5,236	1,875
10-year fracture incidence per 10,000
Hip fractures	0.10	0.11	0.10	−0.009	−0.003
Vertebral fractures	1.07	1.39	1.25	−0.317	−0.181
Wrist fractures	0.03	0.03	0.03	0.005	0.004

Costs expressed in 2018 values.

^aPostmenopausal women age ≥65 years with a femoral neck T-score of ≤−2.5 and history of vertebral fracture.

Table 7. Costs and fracture results for scenario model: total PMO population^a.

	Denosumab	No pharmacologic treatment	Alendronate	Denosumab versus no pharmacologic treatment	Denosumab versus alendronate
Direct fracture-related medical cost (THB)	83,585	91,715	87,991	−8,130	−4,406
Hospitalization cost	22,255	25,074	23,853	−2,819	−1,598
Long-term care cost	23,438	23,948	23,524	−510	−86
Outpatient cost	37,893	42,693	40,614	−4,801	−2,722
Direct fracture-related non-medical and informal care costs (THB)	48,553	54,776	52,051	−6,213	−3,498
Intervention cost (THB)	64,492	0	10,191	64,492	54,301
Drug cost	59,250	0	6,823	59,250	52,427
Treatment management	5,242	0	3,368	5,242	1,874
10-year fracture incidence per 10,000
Hip fractures	0.04	0.05	0.05	−0.005	−0.002
Vertebral fractures	0.75	0.96	0.87	−0.215	−0.122
Wrist fractures	0.04	0.03	0.03	0.003	0.003

Costs expressed in 2018 values.

^aPostmenopausal women with osteoporosis age 60 years or older with a femoral neck T-score of ≤−2.5 and a 23.6% prevalence of vertebral fracture.

Table 8. Incremental cost-effectiveness estimate.

	Denosumab	No treatment	Alendronate	Denosumab versus no pharmacologic treatment	Denosumab versus alendronate
Population: age ≥65 years, T-score ≤−2.5, and with history of vertebral fracture
Total cost	212,695	169,425	170,103	43,270	42,592
LY	14.49	14.07	14.24	0.415	0.248
QALY	9.523	9.161	9.309	0.362	0.214
Cost/ LY	–	–	–	152,674	252,557
Cost/ QALY				119,575	199,186
Population: age ≥60 years, T-score ≤−2.5, and with 23.6% vertebral fracture prevalence
Total cost	196,630	146,481	150,234	50,149	46,396
LY	13.122	12.859	12.968	0.263	0.154
QALY	11.268	10.942	11.078	0.326	0.190
Cost/ LY	–	–	–	190,721	301,290
Cost/ QALY	–	–	–	153,879	243,983

Abbreviations. LY, Life years; QALY, quality-adjusted life years.

Probabilistic sensitivity analysis for the base-case population indicates that denosumab is cost-effective versus no pharmacologic treatment across a range of willingness to pay (WTP) values. With the WTP threshold set at 160,000 THB per QALY, denosumab had a 99.52% and a 9.32% probability of being cost-effective compared with no pharmacologic treatment and to alendronate, respectively (Figure 2(a,b)). One-way sensitivity analysis indicated that the ICER of denosumab versus no treatment was most sensitive to the annual cost of denosumab with a 25% cost reduction lowering the cost per QALY to 78,685 THB (2,361 USD) and a similar increase raising the cost per QALY to 160,466 THB (48,134 USD) (Figure 3(a,b)). Denosumab remained cost-effective versus no treatment with changes to treatment duration (reduced to 2 years; increased to 8 years), 25% change in vertebral fracture costs, and changes in both the excess mortality and costs associated with hip fracture. The ICER of denosumab versus alendronate was also most sensitive to changes in the annual cost of denosumab, followed by excess mortality related to hip fracture. Changes in the annual cost of denosumab, denosumab efficacy for hip fracture, treatment duration, and the first-year cost and utility values for vertebral fracture had similarly small relative effects on the ICER. Changes in the first-year cost and utility values for hip fracture had essentially negligible impact (Figure 3(a,b)). In the base-case population, the number needed to treat (NNT) to avoid one hip fracture in ten years was 106 for denosumab and 165 for alendronate. In the total PMO population, the corresponding numbers needed to treat (NNT) were 214 for denosumab and 345 for alendronate.

Figure 2. Probabilistic sensitivity analysis: (a) denosumab vs. alendronate and no pharmacologic treatment in high-risk patients: (b) denosumab vs. no pharmacologic treatment in high-risk patients. Abbreviation. WTP: willingness to pay. Postmenopausal women age ≥ 65 years with a femoral neck T-score of ≤−2.5 and history of vertebral fracture.

Figure 3. One-way sensitivity analysis: (a) denosumab vs. no pharmacological treatment in high-risk patients: (b) denosumab vs. alendronate in high-risk patients.

Discussion

The results of this analysis demonstrate that denosumab offers a cost-effective treatment option to reduce the risk of fracture in postmenopausal Thai women with osteoporosis. In the high-risk (base-case) population, denosumab achieves superior health outcomes (0.362 and 0.214 QALYs gained versus no pharmacologic treatment and alendronate respectively) and attains ICERs below the cost-effectiveness threshold of 160,000 THB/QALY compared with no treatment. The findings from this study provide important insights into the cost-effectiveness of denosumab versus no treatment and versus alendronate, the most commonly used therapy in Thailand; this information fills an important gap in the literature. In order to provide broadly useful information, the study assessed cost-effectiveness in both an overall osteoporosis population, and in a high-risk population whose characteristics are consistent with the reimbursement strategy for osteoporosis therapy in Thailand³². Women in the high-risk population (i.e. women age 65 years or older with osteoporosis) are very likely to have primary osteoporosis and, therefore, good candidates for pharmacologic therapy.

The results from our models are consistent with previous economic evaluations that have found denosumab to be cost-effective in the treatment of postmenopausal osteoporosis in the United States, Canada, various countries in the European Union, and in JAPAC (Japan and Asia Pacific) countries including Japan, Australia, Taiwan and Singapore^10–17. Moreover, denosumab would also provide a unique and cost-effective option for high-risk PMO patients who are transitioning away from alendronate for any reasons including side effect experience or the inability to consistently take alendronate as directed. Non-persistence significantly blunts the fracture risk reduction effect of alendronate and other oral bisphosphonates, leaving many patients at elevated risk of fracture even after initiating pharmacologic PMO therapy³³. Effective pharmacologic therapy, which relies on good medication adherence (i.e. compliance with dosing instructions and persistence), is particularly important in countries where support services designed to help patients mitigate non-physiological fracture risk factors (e.g. community fall risk assessments and prevention programs) may be lacking.

Bisphosphonates are contraindicated for use in patients with renal impairment and there was evidence showed that bisphosphonates were associated with deterioration in kidney function until acute kidney failure in some cases^34–36. Therefore, routine monitoring of kidney function is advised particularly among patients who are concurrently using other medications that impact kidney function^34–37. The pharmacokinetics and pharmacodynamics of denosumab used at the standard dose, are unaffected by the presence of renal impairment³⁸. In fact, among patients with chronic kidney disease, denosumab provides greater bone mineral density (BMD) gains than bisphosphonates, although vitamin D supplementation is required to prevent hypocalcemia among denosumab users in this population³⁸. Impaired renal function, which becomes more common with advancing age, increases the risk of progressive bone loss and osteoporosis. This association between age and renal impairment underscores the importance of having an osteoporosis treatment option available, for patients who cannot use a bisphosphonate therapy because of either existing renal impairment or their use of potentially nephrotoxic medications to treat comorbid conditions^39–42.

In Thailand, the societal importance of osteoporosis and the impact of osteoporosis-related fractures are growing as the population ages. For instance, the incidence of hip fracture, a costly and debilitating fracture, is increasing⁴³, with postmenopausal women and the elderly at greatest risk. Osteoporosis-related fractures of all types (hip, vertebral, and nonvertebral) exert substantial clinical, economic and humanistic burdens with the impact extending from the patient level up to the societal level. In addition to conferring ongoing healthcare costs and disability, osteoporosis-related fractures also increase the risk of mortality. A study of osteoporotic hip fracture patients treated in Chiang Mai in 2006 and 2007, reported a one-year mortality rate of 21%, which was approximately 9 times that of individuals of the same age in the general population⁴⁴.

There are a number of limitations to the present findings. First, although preference was given to use recent data specific to Thailand whenever possible to create parameter estimates for the model, older data and data from other countries were used when there were gaps in the available data. For example, the age-specific incidence of wrist fracture was obtained from a Japanese paper by Hagino et al. 1999 cited by Thailand Health Intervention and Technology Assessment Program (HITAP) economic evaluation⁴⁵. It is possible that these data obtained from outside of Thailand do not accurately reflect the true experience in Thailand, and their inclusion may add uncertainty to the parameter estimates used in the model. Furthermore, the lack of local data on ‘other osteoporotic fractures’ precluded their inclusion in the model, although this omission actually provides more conservative estimates of the ICERs. Second, the cost data used in this analysis were based on a local study published in 2005. While these data were specific to Thailand, they may be outdated, especially given the technology advancements related to hip replacement surgery. Consequently, the costs of hip replacement may be underestimated, favoring the less efficacious treatment option. However, a one-way sensitivity analysis showed that different assumptions for these costs had little effect on the ICER. Third, the model in this study focused explicitly on hip, vertebral, and wrist fractures, and therefore, provides a conservative estimate of the overall potential impact of osteoporosis-related fractures. In addition, the use of a hierarchy of fracture types in designating transitions between health states, such as allowing post-hip fracture patients to only be at risk for additional hip fractures or death, and allowing women with either a hip or vertebral fracture to no longer be at risk of future wrist fractures may have led to more conservative estimates of the overall fracture burden in the modeled populations. An increased risk of multiple vertebral fractures after discontinuation of denosumab has been observed in data from the FREEDOM trial and its extension (incidence 3.4% among discontinuers and 0.7% among all denosumab users)^46,47 and in a recent real-world study (incidence 0.8% in discontinuers)⁴⁸. However, given the low incidence and current lack of strong evidence⁴⁹, the fracture risk assumption used in the model was not adjusted to reflect potential changes in risk after denosumab treatment discontinuation. Adverse events associated with alendronate were not considered in the model; a decision informed by the cost-effectiveness analysis conducted by Jonsson et al.¹³ in which a sensitivity analysis was done to evaluate the impact of gastrointestinal events associated with alendronate. A minimal impact on the ICER was found (€27,090/QALY base case versus €26,595/QALY including gastrointestinal events). Therefore, adverse events were not included in this analysis. Finally, since the ICER for denosumab in the broader postmenopausal osteoporosis population is relatively close to the WTP threshold of 160,000 THB per QALY, it will be important to re-evaluate cost-effectiveness in this population in the future as the Thai economy continues growing and expanding.

An important strength of this study is that it adheres to the guidance recently provided by Hiligsmann et al. for economic evaluations of osteoporosis therapies including our use of local data to the extent possible to ensure that the model reflects country-specific (i.e. Thailand) osteoporosis management strategies and our detailed presentation of the model parameters and results for specific outcomes (e.g. fracture events, life years, costs per QALY)⁵⁰. Our models examined cost-effectiveness in women with T-score ≤−2.5 at the femoral neck, with separate models provided for a high risk population (mean age of 65 years, a T-score ≤−2.5 at the femoral neck, and a history of vertebral fracture) which served as the base case, and for an overall osteoporosis population consisting of post-menopausal women with mean age of 60 years, a T-score ≤−2.5 at the femoral neck, 23.6% of whom were assumed to have had prior vertebral fracture. The characteristics of this base-case and the inclusion of the additional population with a differing age/risk profile are also consistent with the guidance. In fact, our study adheres to all of the nine items on Hiligsmann et al. checklist, with one exception – the recommendation for reporting transition probabilities. Although, we have not explicitly reported the transition probabilities used in our models, they are described in a previous paper¹³ which we have cited. We have also included a general description of the health state transitions (including a figure [Figure 1]) to illustrate how women may move between states in each model cycle.

In conclusion, the current study demonstrates that, from the societal perspective, denosumab offers a cost-effective PMO treatment option versus no pharmacologic treatment for high-risk postmenopausal women with a T-score ≤ −2.5 at the femoral neck and age ≥65 years. This study offers important data for consideration by stakeholders and decision-makers as they formulate policies to ensure adequate access to appropriate therapies in the face of Thailand’s growing burden of osteoporosis and fracture risk in the future.

Transparency

Declaration of funding

This study was funded by Amgen Thailand Limited.

Declaration of financial interests

The authors have no conflicts of interest to declare. JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Author contributions

Prof. Chatlert Pongchaiyakul: Study design, provided data and review literature, manuscript writing

All other authors contributed to the study design, interpretation of results, and writing of this manuscript.

Acknowledgements

The authors would like to thank Sally Wade (Wade Outcomes Research and Consulting, Salt Lake City, Utah, USA) for medical writing.