Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jul 1.
Published in final edited form as: Osteoporos Int. 2020 Feb 4;31(7):1273–1282. doi: 10.1007/s00198-019-05258-2

Cost-effectiveness of Five Versus Ten Years of Alendronate Treatment Prior to Drug Holiday for Women with Osteoporosis

Smita Nayak 1, Susan L Greenspan 2
PMCID: PMC8215100  NIHMSID: NIHMS1654890  PMID: 32020265

Abstract

Purpose:

We performed a cost-effectiveness analysis to compare 5 versus 10 years of alendronate treatment prior to 5-year drug holiday for postmenopausal osteoporotic women.

Methods:

We created an individual-level state-transition microsimulation model to compare 3 treatment strategies for U.S. postmenopausal women with osteoporosis and femoral neck BMD T-scores between –2.5 and –3.5 at baseline: recurrent periods of 5 years of alendronate followed by 5 years of drug holiday (alendronate 5/5), recurrent periods of 10 years of alendronate followed by 5 years of drug holiday (alendronate 10/5), and no alendronate treatment.

Results:

Base-case analysis revealed for women initiating treatment at ages 50, 60, and 70, the alendronate 5/5 strategy dominated (was more effective and less costly than) the alendronate 10/5 strategy and no treatment. For women age 80, the alendronate 10/5 strategy dominated. When assuming a lower relative risk of nonvertebral fracture during years 6–10 of alendronate treatment than the base-case assumption, the alendronate 10/5 strategy became the most cost-effective strategy even at younger treatment initiation ages. Probabilistic sensitivity analysis results supported the base-case findings: for treatment initiation ages of 50, 60, and 70, the alendronate 5/5 strategy was favored, whereas for treatment initiation age of 80, the alendronate 10/5 strategy was favored; however, there was uncertainty in these findings.

Conclusions:

After 5 years of alendronate treatment, younger postmenopausal women (ages 50–70) with osteoporosis would likely benefit from a drug holiday, whereas older women (age 80) are likely to benefit from treatment for 10 years before a drug holiday.

Keywords: osteoporosis, alendronate, bisphosphonates, drug holiday, cost-effectiveness analysis

Summary

We performed a cost-effectiveness analysis comparing 5 versus 10 years of alendronate treatment prior to 5-year drug holiday for U.S. postmenopausal women with hip BMD T-scores between –2.5 and –3.5. We found that for most postmenopausal women 5 years of treatment prior to drug holiday is the more effective and cost-effective option.

Introduction

Osteoporosis is a common medical condition that affects 10 million adults in the United States including 8 million women [1], and it is estimated that half of postmenopausal women will sustain an osteoporotic fracture in their lifetime [2], with significant associated costs, morbidity, and mortality [1–6]. Bisphosphonates are the most commonly used class of medications to treat osteoporosis and reduce fracture risk, and alendronate is the most commonly prescribed bisphoshponate [7]. Indefinite osteoporosis treatment with bisphosphonates is not recommended due to limited long-term efficacy data and greater risk of rare serious adverse events with longer treatment durations [8]. Additionally, bisphosphonates bind to bone and can remain bound for many years, resulting in residual pharmacological activity for years after discontinuation. As such, periods of treatment discontinuation, or drug holidays, are recommend for patients being treated with bisphopsonates. The American Society for Bone and Mineral Research (ASBMR) and the National Osteoporosis Foundation (NOF) recommend that decisions on when to discontinue bisphosphonates be individualized, with suggestions to consider discontinuing treatment after 3 years of intravenous (IV) bisphosphonate treatment or 5 years of oral bisphosphonate treatment for individuals deemed to not be at high risk for fracture, while perhaps continuing treatment beyond 3 to 5 years for individuals at high risk [8, 9]. In a recent systematic review and meta-analysis on the effect of bisphosphonate drug holidays on bone mineral density (BMD) and osteoporotic fracture risk, we found that women who have low hip BMD T-scores (e.g., ≤ –2.5) after 3–5 years of initial bisphosphonate treatment may benefit from treatment continuation [10]. For alendronate specifically, the evidence from the Fracture Intervention Trial Long-term Extension (FLEX) study showed a reduced risk of clinical vertebral fracture for individuals who continued therapy for 10 years compared to 5 [11].

We performed a cost-effectiveness analysis to compare 5 years versus 10 years of alendronate treatment prior to drug holiday for U.S. postmenopausal women with hip BMD T-scores between –2.5 and –3.5 to help guide treatment duration and drug holiday decisions for women with osteoporosis taking alendronate.

Methods

We created an individual-level state-transition microsimulation model of osteoporosis treatment for community-dwelling U.S. postmenopausal women with femoral neck T-scores between –2.5 and –3.5 at baseline to directly compare 3 treatment strategies: recurrent periods of 5 years of alendronate treatment followed by 5 years of drug holiday (alendronate 5/5), recurrent periods of 10 years of alendronate treatment followed by 5 years of drug holiday (alendronate 10/5), and no alendronate treatment. The model’s perspective is payer, time horizon is lifetime, and estimated outcomes include direct societal costs in 2018 U.S. dollars; quality-adjusted life-years (QALYs); and incremental cost-effectiveness ratios (ICERs). We used TreeAge Pro 2018 software (TreeAge Software, Williamstown, Massachusetts), to perform all analyses.

Model structure and overview

Figure 1 shows a simplified schematic of the model. We constructed an individual-level state transition microsimulation model in a decision tree format using TreeAge Pro 2018 software (TreeAge Software, Williamstown, Massachusetts). The model compared treatment strategies of alendronate 5/5, alendronate 10/5, and no alendronate treatment for U.S. postmenopausal women with osteoporosis with femoral neck T-scores between –2.5 and –3.5 at baseline. Separate analyses were performed for women initiating osteoporosis treatment at age 50, 60, 70, and 80. We assumed that all simulated women were initially community-dwelling, and that baseline vertebral fracture prevalence was 10% for women initiating treatment at age 50; 20% for women initiating treatment at age 60; 30% for women initiating treatment at age 70; and 40% for women initiating treatment at age 80, similar to age-stratified prevalences of vertebral deformity reported in a large cohort study [12]. We assume that all simulated women had no prior nonvertebral fracture history.

Figure 1. Model schematic representation.

Figure 1.

The model compares 3 treatment strategies for postmenopausal women with osteoporosis: recurrent periods of 5 years of alendronate treatment followed by 5 years of drug holiday (alendronate 5/5), recurrent periods of 10 years of alendronate treatment followed by 5 years of drug holiday (alendronate 10/5), and no alendronate treatment. All simulated individuals start the model in either the no fracture or vertebral fracture state, depending on whether they have a history of prior vertebral fracture. Over the remaining lifetime events that may occur include new osteoporotic fractures, alendronate adverse events, moving to a nursing home, and death. Simulated individuals move through the outcomes and fracture state sections of the model on a 3-month cycle. Moph = morphometric, vert = vertebral.

Initial DXA BMD T-score values at the femoral neck for each simulated woman were sampled from a triangular distribution with minimum and maximum limits of –3.5 and –2.5, with T-scores of –2.5 being the mode (the likeliest value). We assumed femoral neck T-scores declined over time for women not on alendronate treatment, with average annual changes derived from published U.S. population-based data [13]. During each 3-month cycle of the model potential events that could occur to each individual included new osteoporotic fractures (hip, morphometric vertebral, clinical vertebral, wrist, or humerus); survival or death; staying in the community or moving to a nursing home; and adverse events of alendronate treatment. During each model cycle only one new osteoporotic fracture at most could occur. Each simulated individual’s age, T-score, and prior fracture history were tracked through the model run, and that individual history affected probability of new fractures, survival or death, and nursing home placement. Mortality risk during each model cycle was affected by age (i.e., age-related risk of mortality due to any health condition) as well as occurrence of hip fracture in the prior year. The occurrence of fractures, treatment-related adverse events, and nursing home residence affected costs and disutility (reduction in health-related quality of life) incurred by each individual. The model is osteoporosis-specific, and thus costs and disutility associated with morbidity from other non-osteoporosis medical conditions were not included in the model. Each individual simulation was continued until the simulated individual died (lifetime time horizon). Average lifetime accrued direct costs and QALYs for each of the strategies were calculated from multiple simulations, which in turn were used to calculate the relative cost-effectiveness of the strategies. Key model parameter assumptions are shown in Table 1.

Table 1.

Key Model Parameter Values

Parameter Base-Case Value Probabilistic Sensitivity Analysis Values (2.5%–97.5%)ab
Fracture related parameters
Relative risk of vertebral fracture, years 1–5 of alendronate treatment 0.52 [14] N/Ac
Relative risk of hip fracture, years 1–5 of alendronate treatment 0.47 [14] N/Ac
Relative risk of nonvertebral fracture, years 1–5 of alendronate treatment 0.64 [14] N/Ac
Relative risk of clinical vertebral fracture, years 6–10 of alendronate treatment 0.57 [11] 0.23–1.40 [11]
Relative risk of morphometric vertebral fracture, years 6–10 of alendronate treatment 0.86 [11] 0.60–1.22 [11]
Relative risk of nonvertebral fracture (including hip), years 6–10 of alendronate treatment 0.77 [11] 0.50–1.20 [11]
Relative risk of any osteoporotic fracture per standard deviation decrease in femoral neck bone mineral density 1.27 [19] 1.14–1.40
Relative risk of fracture (any site) with prior vertebral fracture 1.86d [19] 1.67–2.04
Proportion of all vertebral fractures that are clinically detected 0.226 [20] N/Ac
Costs, 2018 US dollarse
Alendronate average annual costf 200 [26] 160–240
Hip fracture, direct medical costs 38,746 [27] 30,997–46,495
Clinical vertebral fracture, direct medical costs 11,797 [27] 9,438–14,156
Wrist fracture, direct medical costs 8,674 [27] 6,939–10,409
Humerus fracture, direct medical costsg 8,674 [27] 6,939–10,409
Nursing home care, annual cost 89,297 [28] 71,438–107,156
Subtrochanteric femoral fracture 38,746 [27] 30,997–46,495
Esophageal ulcerh 4,034 [29, 30] 3,227–4841
Osteonecrosis of the jawi 1,000 800–1,200
Esophagitisij 314 [26, 30] 251–377
Health state utility values
No fracture, age 45–54 years 0.87 [21] 0.83–0.91
No fracture, age 55–64 years 0.84 [21] 0.80–0.88
No fracture, age 65–74 years 0.84 [21] 0.80–0.88
No fracture, age ≥ 75 0.82 [21] 0.78–0.86
Hip fracture, first year/subsequent years (multiplier) 0.797/0.899 [22] 0.757–0.837/0.854–0.944
Vertebral fracture, first year/subsequent years (multiplier) 0.860/0.965 [22] 0.817–0.903/N/Ac
Humerus fracture, first year/subsequent years (multiplier) 0.91/1.00 [22] 0.865–0.956/N/Ac
Wrist fracture, first year/subsequent years (multiplier) 0.94/1.00 [22] 0.893–0.987/N/Ac
Nursing home residence (multiplier) 0.40 [23] 0.30–0.50
Subtrochanteric femoral fracturek, first year/subsequent years (multiplier) 0.797/.899 [22] 0.757–0.837/0.854–0.944
Osteonecrosis of the jaw (disutility) 0.13 [24] N/Ac
Esophageal ulcerl (disutility) 0.08 [24] N/Ac
Esophagitisl (disutility) 0.08 [24] N/Ac
Risk of nursing home entry after hip fracture
Relative risk of nursing home admission in year after hip fracture 3.96 [34] 3.56–4.36
Hip fracture mortality
In-hospital (Age <75/ 75–84/ ≥85) 0.017/.031/.040 [37] N/Ac
During first year after hip fracture (Ages <70/ 70–79/ ≥80) 0.163/0.165/.203 [38] 0.130–0.196/0.132–0.198/ 0.162–.244
Discount rates
Costs 0.03 [25] N/Ac
Quality-adjusted life-years 0.03 [25] N/Ac
a

Normal distributions assumed

b

Values estimated unless otherwise specified

c

Not Applicable

d

Calculated from data presented in Table 2 of Krege et al. paper

e

Costs not presented in 2018 dollars were inflated to 2018 dollars using the Consumer Price Index for Medical Care through November 2018

f

A midrange price from online pharmacies was assumed

g

Assumed same cost as wrist fracture

h

Cost includes hospitalization for diagnosis-related group (DRG) code 379 (gastrointestinal hemorrhage without complication or comorbidity) plus 2 physician visits (Current Procedural Terminology (CPT) code 99213)

i

Cost assumed

j

Cost includes 1 year of treatment with a proton-pump inhibitor (omeprazole) plus 1 physician visit (CPT code 99213)

k

Utility values for hip fracture assumed

l

Surrogate disutility value for dyspepsia used

Treatment

Three different treatment strategies were compared: recurrent periods of 5 years of alendronate treatment followed by 5 years of drug holiday (alendronate 5/5); recurrent periods of 10 years of alendronate treatment followed by 5 years of drug holiday (alendronate 10/5); and no alendronate treatment. For women receiving alendronate treatment, we assumed fracture risk reduction efficacy in the first 5 years of each treatment period as reported in The Fracture Intervention (FIT) Trials [14], and for women receiving 10 years of alendronate treatment we assumed fracture risk reduction efficacy in years 6 through 10 of each treatment period relative to individuals who discontinued treatment after 5 years as reported in The Fracture Intervention Trial Long-term Extension (FLEX) study [11] (Table 1). For years 6 through 10 of alendronate therapy we used point estimates of fracture risk reduction efficacy for nonvertebral and clinical vertebral fractures obtained from the FLEX subgroup with baseline femoral neck BMD T-scores ≤ –2.5, and for morphometric vertebral fracture risk reduction efficacy we used the point estimate presented for all study participants [11]. We assumed that alendronate fracture risk reduction efficacy declined linearly over each 5-year drug holiday period to no residual benefit [15]. We assumed all women, whether they received alendronate treatment or not, received adequate vitamin D and calcium supplementation.

Fracture rates

Population rates of osteoporotic fractures of the hip, vertebra, humerus, and forearm for U.S. women of various ages were obtained from a paper by Ettinger and colleagues [16], and adjusted for each simulated individual according to their femoral neck BMD Z-score and the existence of a prior vertebral fracture [17–19]. To calculate morphometric fracture rates from the published data on clinical vertebral fracture rates, we assumed that 22.6% of all vertebral fractures were clinically recognized [20]. We assumed that any prior vertebral fracture, clinical or morphometric, increased future fracture risk by a relative risk of 1.86; this value was calculated from data presented in a paper by Krege and colleagues [19].

Health-state utilities

Baseline health-state utility values for women of various ages were obtained from a study reporting nationally representative values [21]. A reduction in health-related quality of life, or “disutility”, occurred in the event of fractures, nursing home stay, and treatment-related adverse events; disutility associated with these events was obtained from several published sources [22–24]. To model the reduction in health-state utility associated with fractures and nursing home residence we used multipliers in which we multiplied baseline utility by the multiplier values shown in the table to arrive at the new utility value [22, 23]; and to model disutility associated with adverse events we subtracted the values shown in the table from the utility value prior to the adverse event [24]. We discounted future utilities at 3% annually [25].

Costs

Direct costs associated with alendronate treatment, osteoporotic fractures, nursing home stay, and treatment-related adverse events were obtained from several different sources and incorporated into the model [26–30]. We discounted future costs at 3% annually [25].

Adverse events

Potential adverse events associated with alendronate treatment, including atypical subtrochanteric femoral fractures, osteonecrosis of the jaw, esophageal ulcer, and esophagitis were incorporated into the model, with rates obtained from published data [31, 32]. We assumed that women who experienced esophagitis initiated treatment with a proton-pump inhibitor and continued alendronate, whereas women who experienced any of the other adverse events discontinued alendronate.

Nursing home admission

Nursing home admission rates and continuance/length of stay for women of different ages and increased relative risk of nursing home entry after hip fracture were obtained from published data [33, 34].

Mortality rates

Mortality rates for women ages 50 through 100 were obtained from 2015 U.S. National Vital Statistics data for all females [35]; and mortality rates for women over age 100 were obtained from Social Security actuarial data [36]. Increased mortality after hip fracture was incorporated into the model [37, 38]; no increased mortality was assumed after other types of fractures in base-case analysis.

Analyses

For each treatment initiation age evaluated (50, 60, 70, and 80) we performed base-case analysis, deterministic one-way sensitivity analyses, and probabilistic sensitivity analyses. Variables evaluated with one-way sensitivity analyses included fracture rates, costs, relative risk of nonvertebral fracture during years 6–10 of alendronate treatment, and risk of mortality after vertebral fracture. Probabilistic sensitivity analyses were conducted to assess the influence of joint input parameter uncertainty on model results. Each base-case analysis and one-way sensitivity analysis was performed with 2 million trials, simulating treatment of 2 million individual women with each of the evaluated strategies. We performed probabilistic sensitivity analysis with 1,000 trials and 5,000 samples, and identified the probability of the evaluated strategies being most cost-effective at willingness-to-pay thresholds of $50,000 per QALY and $100,000 per QALY [39].

Model validation

We compared model estimates for life expectancy and fractures with published data to validate the model [5, 35, 40].

Results

Model validation

The model predicted median life expectancies for 50, 60, 70, and 80-year-old women with osteoporosis and baseline femoral neck T-scores between –2.5 and –3.5 not receiving alendronate treatment of 84.0, 84.8, 86.0, and 88.8, respectively, similar to the 2015 National Vital Statistics life expectancy figures for all women of 83.3, 84.6, 86.6, and 89.7 years, respectively [35]. The model predicted that 30.5% of 70-year-old women and 28.2% of 80-year-older women with osteoporosis who did not receive alendronate treatment would experience a hip fracture over their remaining lifetimes, similar to 25-year probabilities of hip fracture reported by Black et al. for women in the Study of Osteoporotic Fractures (SOF) with a femoral neck BMD T-score of ≤ –2.5 who were age 70–74 and 75 and older at baseline of 32.5% and 31.5%, respectively [40]. The model predicted that 48.8% of 70-year-old women with osteoporosis who did not receive alendronate treatment would experience a nonvertebral fracture (hip, humerus, or wrist) over their remaining lifetimes, similar to a 20-year probability of nonvertebral fracture of 46.2% reported by Black et al. for women in SOF who were age 70–74 at baseline [40]. The model predicted that 60.6% of 60-year-old women with osteoporosis not receiving alendronate treatment would experience a hip, humerus, wrist, or clinical vertebral over their remaining lifetimes, similar to a mortality-adjusted residual lifetime risk of osteoporotic fracture of 64.6% reported for a population of white women aged 60 and older with femoral neck BMD T-scores of ≤–2.5 [5].

Base-case analyses

Base-case analysis results for U.S. postmenopausal women with osteoporosis and femoral neck BMD T-scores between –2.5 and –3.5 at baseline for each of the treatment initiation ages evaluated (50, 60, 70, and 80) are shown in Table 2. For treatment initiation ages of 50, 60, and 70, the treatment strategy of recurrent periods of 5 years of alendronate treatment followed by 5-year drug holiday periods (alendronate 5/5) dominated (was more effective as measured by QALYs and less costly than) the strategies of recurrent periods of 10 years of alendronate treatment followed by 5-year drug holiday periods (alendronate 10/5) and no alendronate treatment. For the treatment initiation age of 80, the treatment strategy of alendronate 10/5 dominated the strategies of alendronate 5/5 and no alendronate treatment. No alendronate treatment was significantly more expensive and less effective than either of the alendronate treatment strategies at all evaluated treatment initiation ages.

Table 2.

Base-Case Analysis Results By Treatment Initiation Age

Treatment Strategy Lifetime Cost ($)a Incremental Cost ($)a Quality-adjusted life-years (QALYs) accrued Incremental QALYs Incremental Cost-Effectiveness Ratio (ICER)b
Age 50
Alendronate 5/5 28,627 – 16.6723 – dominantc
Alendonate 10/5 29,172 545 16.6665 −0.0058 dominatedd
No alendronate treatment 31,291 2,664 16.5200 −0.1524 dominatedd
Age 60
Alendronate 5/5 36,102 – 13.3573 – dominantc
Alendonate 10/5 36,772 671 13.3465 −0.0108 dominatedd
No alendronate treatment 39,896 3,794 13.1870 −0.1703 dominatedd
Age 70
Alendronate 5/5 40,692 – 9.6910 – dominantc
Alendonate 10/5 40,893 201 9.6861 −0.0049 dominatedd
No alendronate treatment 45,462 4,770 9.5098 −0.1812 dominatedd
Age 80
Alendronate 10/5 29,489 – 6.1057 – dominantc
Alendonate 5/5 29,509 21 6.0960 −0.0097 dominatedd
No alendronate treatment 34,825 5.9555 −0.1502 dominatedd
a

Costs in 2018 U.S. dollars

b

Incremental cost-effectiveness ratios are the cost per quality-adjusted life-year gained for a strategy compared to the next less costly nondominated strategy. ICERs are not shown for strategies that are dominante or dominatedd.

c

Dominant strategies are less expensive and more effective than other strategies

d

Dominated strategies are more expensive and less effective than another strategy

One-way sensitivity analyses

One-way sensitivity analysis results when varying costs, incident fracture rates (hip, vertebral, humerus, and forearm), relative risk of nonvertebral fracture during years 6–10 of alendronate treatment, and risk of mortality after vertebral fracture are shown in Supplemental Table 1. Results for one-way sensitivity analyses in which all fracture rates were varied 20% lower or higher and costs were varied 20% lower or higher than base-case estimates were similar to base-case analysis findings: in these sensitivity analyses, for treatment initiation ages of 50, 60, or 70, the alendronate 5/5 treatment strategy remained most cost-effective, and for a treatment initiation age of 80 the alendronate 10/5 treatment strategy remained most cost-effective. Likewise, results for one-way sensitivity analyses in which we assumed a 10% or 20% increased relative risk of mortality for women with any prior vertebral fracture (clinical or morphometric) for the remainder of her lifetime were similar to base-case analysis findings. However, results were more sensitive to assumptions about relative risk of nonvertebral frature during years 6–10 of alendronate treatment. When assuming a relative risk of nonvertebral fracture of 0.70 during years 6–10 of alendronate treatment (instead of the base-case assumption of a relative risk of nonvertebral fracture of 0.77 during years 6–10 of alendronate treatment), the alendronate 10/5 treatment strategy became the most cost-effective treatment strategy for women initiating treatment at age 70 in addition to women initiating treatment at age 80; the alendronate 5/5 treatment strategy remained most cost-effective for women initiating treatment at ages 50 or 60. When assuming a relative risk of nonvertebral fracture of 0.60 during years 6–10 of alendronate treatment, the alendronate 10/5 treatment strategy became the most cost-effective treatment strategy for women initiating treatment at all ages evaluated (50, 60, 70, and 80). Conversely, when assuming a relative risk of nonvertebral fracture of 0.90 during years 6–10 of alendronate treatment, the alendronate 5/5 treatment strategy became the most cost-effective treatment strategy for women initiating treatment at all ages evaluated, including age 80.

Probabilistic sensitivity analyses

Probabilistic sensitivity analysis results are shown in Figure 2. For treatment initiation ages of 50, 60, and 70, the alendronate 5/5 treatment strategy was favored in a majority of the iterations for willingness-to-pay thresholds of either $50,000/QALY or $100,000/QALY, whereas for a treatment initiation age of 80 the alendronate 10/5 treatment strategy was favored in a majority of the iterations. For each treatment initiation age evaluated, neither the alendronate 5/5 nor the alendronate 10/5 strategy was favored in greater than two-thirds of the probabilistic sensitivity analysis iterations for willingness-to-pay thresholds of $50,000/QALY or $100,000/QALY; the other alendronate treatment strategy was favored in at least one-third of the iterations for each treatment initiation age evaluated.

Figure 2. Probabilistic sensitivity analysis cost-effectiveness acceptability curves for treatment initiation ages of 50 (a), 60 (b), 70 (c), and 80 (d).

Figure 2.

The y-axis shows the proportion of iterations that each evaluated strategy is favored over the range of willingness-to-pay thresholds (in 2018 dollars per quality-adjusted life-year) shown on the x-axis.

Discussion

Our cost-effectiveness analysis comparing 5 years versus 10 years of alendronate treatment prior to a 5-year drug holiday for U.S. postmenopausal women with hip BMD T-scores between –2.5 and –3.5 at baseline found that for younger postmenopausal women such as those initiating treatment at ages of 50, 60, or 70 a treatment strategy of recurrent periods of 5 years of alendronate treatment followed by 5 year drug holiday periods (alendronate 5/5) is most likely to be most cost-effective, whereas for older postmenopausal women such as those initiating treatment at age 80 a treatment strategy of recurrent periods of 10 years of alendronate treatment prior to 5 year drug holiday periods (alendronate 10/5) is more likely to be most cost-effective. One-way sensitivity analyses revealed that our findings are sensitive to assumptions about nonvertebral fracture risk reduction during years 6–10 of alendronate treatment; if we assume nonvertebral fracture relative risk estimates lower than the point estimate from the FLEX study, 10 years of alendronate treatment prior to drug holiday becomes more cost-effective even for younger postmenopausal women, whereas if we assume nonvertebral fracture relative risk estimates higher than the point estimate from the FLEX study, 5 years of alendronate treatment prior to drug holiday becomes the more cost-effective option even for older postmenopausal women (e.g., age 80) [11]. Probabilistic sensitivity analyses results for willingness-to-pay thresholds of $50,000/QALY and $100,000/QALY supported the base-case analysis findings: for treatment initiation ages of 50, 60, and 70, the alendronate 5/5 treatment strategy was favored in a majority of the iterations, whereas for a treatment initiation age of 80 the alendronate 10/5 treatment strategy was favored in a majority of the iterations. However, neither the alendronate 5/5 nor the alendronate 10/5 strategy was favored in greater than two-thirds of the probabilistic sensitivity analysis iterations for willingness-to-pay thresholds of $50,000/QALY or $100,000/QALY for any of the treatment initiation ages evaluated; the other alendronate treatment strategy was favored in at least one-third of the iterations for each treatment initiation age evaluated.

Our findings suggest that for younger postmenopausal women (e.g., women age 50–70 years) with osteoporosis and hip BMD T-scores between –2.5 and –3.5 after 5 years of initial treatment with alendronate, it is likely more effective and cost-effective to discontinue alendronate for 5 years (take a 5-year drug holiday) rather than to continue treatment for a total of 10 years before taking a 5-year drug holiday. This finding is predicated on the assumption that these women would restart alendronate treatment for recurrent 5-year treatment periods followed by 5-year drug holidays as they get older. For older postmenopausal women (e.g., age 80 years) with osteoporosis and hip BMD T-scores between –2.5 and –3.5 after 5 years of initial treatment with alendronate, our findings suggest that it is likely more effective and cost-effective to continue alendronate for a total of 10 years before taking a 5-year drug holiday, rather than discontinuing alendronate treatment after 5 years. Probabilistic sensitivity analyses revealed that there is significant uncertainty in these findings; however, given our best estimates for potential ranges of key model parameters at this time, it is more likely that a recurrent 5-year on/5-year off treatment strategy with alendronate is the better option for younger postmenopausal women, while a recurrent 10-year on/5-year off alendronate treatment strategy may be better for older women. For women initiating osteoporosis treatment in their 70s (age 71–79 years), our analyses did not provide an answer as to which alendronate treatment strategy is likely more effective and cost-effective. Thus, for women in their 70s we recommend a case-by-case approach according to patient preference; however, if no preference is clear a reasonable recommendation may be to treat women in their early 70s with the alendronate 5/5 strategy (similar to treatment suggestion for younger women) and women in their late 70s with the alendronate 10/5 strategy (similar to treatment suggestion for older women).

The American Society for Bone and Mineral Research (ASBMR) and the National Osteoporosis Foundation (NOF) currently recommend that decisions on when to discontinue bisphosphonates be individualized, with suggestions to consider discontinuing oral bisphosphonate treatment after 5 years for individuals deemed to not be at high risk for fracture, while perhaps continuing treatment beyond 5 years for individuals at high risk [8, 9]. Specifically, the NOF recommends that a thorough risk assessment be performed after an initial 3 to 5 year bisphosphonate treatment period, and individuals at high fracture risk should continue bisphosphonate treatment or another therapy, while it is reasonable to discontinue bisphosphonates for those with modest fracture risk [8]. An ASBMR task force recommended that after 5 years of oral bisphosphonate treatment risk should be reassessed, and for women with a prior major osteoporotic fracture or fracture on therapy, a low hip T-score, high fracture risk score, or older women, continuation of bisphosphonate therapy for up to 10 years should be considered [9]. Our cost-effectiveness analysis findings suggest that younger postmenopausal women (e.g. ages 50–70) with osteoporosis with hip BMD T-scores between –2.5 and –3.5 after 5 years of initial treatment with alendronate would likely benefit from treatment discontinuation for 5 years instead of continuing therapy for 10 years, whereas older women (e.g. age 80) with low hip BMD T-scores between –2.5 and 3.5 after 5 years of initial treatment are more likely to benefit from treatment up to 10 years before discontinuation for 5 years. Thus, our findings differ somewhat from the NOF and ASBMR recommendations in that they suggest younger postmenopausal women, even those at higher risk for fracture based on a hip BMD T-score ≤ –2.5 after 5 years of initial treatment with alendronate, are more likely to benefit from a drug holiday after 5 years of treatment rather than continuing treatment for 10 years prior to taking a holiday.

This study is the first cost-effectiveness analysis that we are aware of that has compared different durations of bisphosphonate treatment prior to drug holiday to inform this key treatment decision for individuals with osteoporosis. It is unlikely that a clinical trial with lifetime follow-up will be conducted to compare different bisphosphonate treatment durations prior to drug holiday, and thus a modeling analysis such as this one is essential to help inform the important clinical question of how long to treat individuals with osteoporosis prior to drug holiday. However, this study has several limitations that are important to note. First, as in any modeling analysis, there is uncertainty in the estimates used for parameter values, as well as uncertainty in the model structure with respect to whether all key factors are included. The validity and generalizability of our findings is dependent on the quality of our data sources; including the use of single data sources for some key model parameter values. Despite performing sensitivity analyses to test the impact of parameter value uncertainty on our findings, it is still possible that there are biases in the parameter values and ranges we assumed and/or biases inherent in the structure of our model. Second, our analyses only included four common types of osteoporotic fractures (vertebral, wrist, humerus, and hip), not all potential osteoporotic fractures, due to a dearth of high-quality incidence data for other types of osteoporotic fractures. Third, we assumed that after a drug holiday period of 5 years, women who restart alendronate therapy would experience treatment effectiveness in terms of fracture risk reduction equivalent to treatment-naïve women. However, because bisphosphonates are long lasting, this may not be the case. Fourth, our analyses are specific for U.S. postmenopausal women with femoral neck T-scores between –2.5 and –3.5 at baseline, and our findings may not apply to other women. Furthermore, we assumed equivalent alendronate efficacy for fracture risk reduction in years 6–10 of treatment for women with low hip BMD T-scores who did and did not have a prior osteoporotic fracture; however, it is possible that treatment efficacy during years 6–10 of alendronate treatment, and thus cost-effectiveness of 5 versus 10 years of alendronate treatment prior to drug holiday, may vary depending on prior fracture status. Additionally, we assumed a fixed drug holiday length of 5 years, rather than evaluating the impact of different drug holiday lengths on our findings for the relative cost-effectiveness of 5 versus 10 years of alendronate treatment prior to drug holiday. Moreover, we only included women who were initially community-dwelling. Older women who live in long-term care facilities have a higher risk of fracture, more adverse events, and would require a different cost-effectiveness model. Finally, we only included a single oral bisphosphonate, alendronate. Our findings may not apply to other oral or intravenous bisphosphonates.

In conclusion, our findings suggest that younger postmenopausal women (e.g. ages 50–70) with osteoporosis with hip BMD T-scores between –2.5 and –3.5 after 5 years of initial treatment with alendronate would likely benefit from treatment discontinuation for 5 years instead of continuing therapy for 10 years, whereas older women (e.g. age 80) with hip BMD T-scores between –2.5 and –3.5 after 5 years of initial treatment are more likely to benefit from treatment for 10 years before discontinuation for 5 years. There is uncertainty in these findings; however, given our best estimates for key model parameters at this time, these are the strategies that we found to be most likely to be most cost-effective. Future research is needed to further clarify key parameter values such as the relative risk of nonvertebral fracture during years 6–10 of alendronate treatment, and to examine other oral and intravenous bisphosphonates.

Supplementary Material

1654890_Supp_Tab1

Acknowledgements

Sources of funding:

Drs. Nayak and Greenspan were supported by grant R21AR072930 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases. Dr. Greenspan was also supported by NIH grants P30AG024827, K07AG052668, and R01AG050302 from the National Institute on Aging.

Footnotes

Conflict of Interest:

Smita Nayak declares no conflict of interest. Susan L. Greenspan receives research funding from Amgen and Radius.

Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of a an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.

References

  • 1.Wright NC, Looker AC, Saag KG, Curtis JR, Delzell ES, Randall S, Dawson-Hughes B (2014) The recent prevalence of osteoporosis and low bone mass in the United States based on bone mineral density at the femoral neck or lumbar spine. J Bone Miner Res 29:2520–2526 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.U.S. Department of Health and Human Services. Bone Health and Osteoporosis: A Report of the Surgeon General. Rockville, MD: U.S. Department of Health and Human Services, Office of the Surgeon General, 2004. [Google Scholar]
  • 3.Osteoporosis prevention, diagnosis, and therapy. NIH Consens Statement. 2000;17(1):1–45 [PubMed] [Google Scholar]
  • 4.Lin JT, Lane JM (2004) Osteoporosis: a review. Clin Orthop Relat Res 126–134 [PubMed] [Google Scholar]
  • 5.Nguyen ND, Ahlborg HG, Center JR, Eisman JA, Nguyen TV (2007) Residual lifetime risk of fractures in women and men. J Bone Miner Res 22:781–788 [DOI] [PubMed] [Google Scholar]
  • 6.Blume SW, Curtis JR (2011) Medical costs of osteoporosis in the elderly Medicare population. Osteoporos Int 22:1835–1844 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Yang X, Sajjan S, Modi A (2016) High rate of non-treatment among osteoporotic women enrolled in a US Medicare plan. Curr Med Res Opin 32:1849–1856 [DOI] [PubMed] [Google Scholar]
  • 8.Cosman F, de Beur SJ, LeBoff MS, Lewiecki EM, Tanner B, Randall S, Lindsay R; National Osteoporosis Foundation (2014) Clinician’s Guide to Prevention and Treatment of Osteoporosis. Osteoporos Int 25:2359–2381 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Adler RA, El-Hajj Fuleihan G, Bauer DC, Camacho PM, Clarke BL, Clines GA, Compston JE, Drake MT, Edwards BJ, Favus MJ, Greenspan SL, McKinney R Jr, Pignolo RJ, Sellmeyer DE (2016) Managing Osteoporosis in Patients on Long-Term Bisphosphonate Treatment: Report of a Task Force of the American Society for Bone and Mineral Research. J Bone Miner Res 31:16–35 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Nayak S, Greenspan SL (2019) A systematic review and meta-analysis of the effect of bisphosphonate drug holidays on bone mineral density and osteoporotic fracture risk. Osteoporos Int 30:705–720 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Black DM, Schwartz AV, Ensrud KE, Cauley JA, Levis S, Quandt SA, Satterfield S, Wallace RB, Bauer DC, Palermo L, Wehren LE, Lombardi A, Santora AC, Cummings SR; FLEX Research Group (2006) Effects of continuing or stopping alendronate after 5 years of treatment: the Fracture Intervention Trial Long-term Extension (FLEX): a randomized trial. JAMA 296:2927–2938 [DOI] [PubMed] [Google Scholar]
  • 12.Jackson SA, Tenenhouse A, Robertson L (2000) Vertebral fracture definition from population-based data: preliminary results from the Canadian Multicenter Osteoporosis Study (CaMos). Osteoporos Int 11:680–687 [DOI] [PubMed] [Google Scholar]
  • 13.Looker AC, Borrud LG, Hughes JP, Fan B, Shepherd JA, Melton LJ 3rd, (2012) Lumbar spine and proximal femur bone mineral density, bone mineral content, and bone area: United States, 2005–2008. Vital Health Stat 11 251:1–132 [PubMed] [Google Scholar]
  • 14.Black DM, Thompson DE, Bauer DC, Ensrud K, Musliner T, Hochberg MC, Nevitt MC, Suryawanshi S, Cummings SR; Fracture Intervention Trial (2000) Fracture risk reduction with alendronate in women with osteoporosis: the Fracture Intervention Trial. FIT Research Group. J Clin Endocrinol Metab 85:4118–4124 [DOI] [PubMed] [Google Scholar]
  • 15.Tosteson AN, Jonsson B, Grima DT, O’Brien BJ, Black DM, Adachi JD (2001) Challenges for model-based economic evaluations of postmenopausal osteoporosis interventions. Osteoporos Int 12:849–857 [DOI] [PubMed] [Google Scholar]
  • 16.Ettinger B, Black DM, Dawson-Hughes B, Pressman AR, Melton LJ 3rd, (2010) Updated fracture incidence rates for the US version of FRAX. Osteoporos Int 21:25–33 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Delmas P, Elsman JA, Fujiwara S, Kroger H, Mellstrom D, Meunier PJ, Melton LJ 3rd, O’Neill T, Pols H, Reeve J, Silman A, Tenenhouse A (2005) Predictive value of BMD for hip and other fractures. J Bone Miner Res 20:1185–1194 [DOI] [PubMed] [Google Scholar]
  • 18.Cummings SR, Cawthon PM, Ensrud KE, Cauley JA, Fink HA, Orwoll ES (2006) BMD and risk of hip and nonvertebral fractures in older men: a prospective study and comparison with older women. J Bone Miner Res 21:1550–1556 [DOI] [PubMed] [Google Scholar]
  • 19.Krege JH, Wan X, Lentle BC, Berger C, Langsetmo L, Adachi JD, Prior JC, Tenenhouse A, Brown JP, Kreiger N, Olszynski WP, Josse RG, Goltzman D; CaMos Research Group (2013) Fracture risk prediction: importance of age, BMD and spine fracture status. Bonekey Rep 2:404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Fink HA, Milavetz DL, Palermo L, Nevitt MC, Cauley JA, Genant HK, Black DM, Ensrud KE (2005) What proportion of incident radiographic vertebral deformities is clinically diagnosed and vice versa? J Bone Miner Res 20:1216–1222 [DOI] [PubMed] [Google Scholar]
  • 21.Fryback DG, Dunham NC, Palta M, Hanmer J, Buechner J, Cherepanov D, Herrington SA, Hays RD, Kaplan RM, Ganiats TG, Feeny D, Kind P (2007) US norms for six generic health-related quality-of-life indexes from the National Health Measurement study. Med Care 45:1162–1170 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Hiligsmann M, Ethgen O, Richy F, Reginster JY (2008) Utility values associated with osteoporotic fracture: a systematic review of the literature. Calcif Tissue Int 82:288–292 [DOI] [PubMed] [Google Scholar]
  • 23.Brazier JE, Green C, Kanis JA (2002) A systematic review of health state utility values for osteoporosis-related conditions. Osteoporos Int 13:768–776 [DOI] [PubMed] [Google Scholar]
  • 24.Solomon DH, Patrick AR, Schousboe J, Losina E (2014) The potential economic benefits of improved postfracture care: a cost-effectiveness analysis of a fracture liaison service in the US health-care system. J Bone Miner Res 29:1667–1674 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Sanders GD, Neumann PJ, Basu A, Brock DW, Feeny D, Krahn M, Kuntz KM, Meltzer DO, Owens DK, Prosser LA, Salomon JA, Sculpher MJ, Trikalinos TA, Russell LB, Siegel JE, Ganiats TG (2016) Recommendations for Conduct, Methodological Practices, and Reporting of Cost-effectiveness Analyses: Second Panel on Cost-Effectiveness in Health and Medicine. JAMA 316:1093–1103 [DOI] [PubMed] [Google Scholar]
  • 26.PharmacyChecker.com. Available at: http://pharmacychecker.com
  • 27.Kilgore ML, Curtis JR, Delzell E, Becker DJ, Arora T, Saag KG, Morrisey MA (2013) A close examination of healthcare expenditures related to fractures. J Bone Miner Res 28:816–820 [DOI] [PubMed] [Google Scholar]
  • 28.Genworth 2018 Cost of Care Survey. Available at https://www.genworth.com/aging-and-you/finances/cost-of-care.html
  • 29.Boston Scientific 2018 GI Procedural Reimbursement Guide. Available at: https://www.bostonscientific.com
  • 30.Centers for Medicare and Medicaid Services. National Physician Fee Schedule. Available at: https://www.cms.gov/apps/physician-fee-schedule/overview.aspx
  • 31.Brown JP, Morin S, Leslie W, Papaioannou A, Cheung AM, Davison KS, Goltzman D, Hanley DA, Hodsman A, Josse R, Jovaisas A, Juby A, Kaiser S, Karaplis A, Kendler D, Khan A, Ngui D, Olszynski W, Ste-Marie LG, Adachi J. (2014) Bisphosphonates for treatment of osteoporosis: expected benefits, potential harms, and drug holidays. Can Fam Physician 60:324–333 [PMC free article] [PubMed] [Google Scholar]
  • 32.Black DM, Cummings SR, Karpf DB, Cauley JA, Thompson DE, Nevitt MC, Bauer DC, Genant HK, Haskell WL, Marcus R, Ott SM, Torner JC, Quandt SA, Reiss TF, Ensrud KE (1996) Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Fracture Intervention Trial Research Group. Lancet 348:1535–1541 [DOI] [PubMed] [Google Scholar]
  • 33.Society of Actuaries Long-Term Care Experience Committee Intercompany Study Report 6 1984–2007. Available at: https://www.soa.org/globalassets/assets/files/research/exp-study/research-ltc-study-1984-report.pdf
  • 34.Tajeu GS, Delzell E, Smith W, Arora T, Curtis JR, Saag KG, Morrisey MA, Yun H, Kilgore ML (2014) Death, debility, and destitution following hip fracture. J Gerontol A Biol Sci Med Sci 69:346–353 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Arias E, Xu J (2018) United States Life Tables, 2015. Natl Vital Stat Rep 67:1–64 [PubMed] [Google Scholar]
  • 36.Social Security Administration Actuarial Life Table. Available at: https://www.ssa.gov/oact/STATS/table4c6.html
  • 37.Orces CH (2013) In-hospital hip fracture mortality trends in older adults: the National Hospital Discharge Survey, 1988–2007. J Am Geriatr Soc 61:2248–2249 [DOI] [PubMed] [Google Scholar]
  • 38.LeBlanc ES, Hillier TA, Pedula KL, Rizzo JH, Cawthon PM, Fink HA, Cauley JA, Bauer DC, Black DM, Cummings SR, Browner WS (2011) Hip fracture and increased short-term but not long-term mortality in healthy older women. Arch Intern Med 171:1831–1837 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Shiroiwa T, Sung YK, Fukuda T, Lang HC, Bae SC, Tsutani K (2010) International survey on willingness-to-pay (WTP) for one additional QALY gained: what is the threshold of cost effectiveness? Health Econ 19:422–437 [DOI] [PubMed] [Google Scholar]
  • 40.Black DM, Cauley JA, Wagman R, Ensrud K, Fink HA, Hillier TA, Lui LY, Cummings SR, Schousboe JT, Napoli N (2018) The Ability of a Single BMD and Fracture History Assessment to Predict Fracture Over 25 Years in Postmenopausal Women: The Study of Osteoporotic Fractures. J Bone Miner Res 33:389–395 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1654890_Supp_Tab1

RESOURCES