Cost-Effectiveness of Alendronate Therapy for Osteopenic Postmenopausal Women

Context Many postmenopausal women have osteopenia but not osteoporosis or fracture history. Information about the cost-effectiveness of treating such women with alendronate should guide clinical recommendations. Contribution The investigators estimate that the costs of treating postmenopausal women with osteopenia (femoral neck T-scores > 2.5) for 5 years with alendronate range from $70000 to $332000 per quality-adjusted life-year. Cautions Unless the cost of alendronate decreases or new data show that alendronate reduces fracture risk more than is currently thought, treatment of osteopenia with alendronate costs more than Americans are typically willing to pay for health care interventions. The Editors Osteoporosis is associated with an increased risk for fractures, which, in turn, is associated with clinically significant morbidity. The broad consensus is that patients who present with fractures should be treated for osteoporosis (1-5). In the absence of fractures, however, the indications for antiresorptive drug therapy are controversial. Many agree that patients with osteoporosis who meet the World Health Organization criteria (bone mineral density [BMD] > 2.5 SDs less than the young healthy mean or a T-score 2.5 [6]) should be treated (1-5). However, many fractures among postmenopausal women occur in those with T-scores better than 2.5 because more women have scores in this range than in the osteoporotic range (7, 8). While some do not recommend treating osteopenic postmenopausal women who do not have a history of clinical fracture (1, 2), guidelines published by the National Osteoporosis Foundation (3) and the American College of Obstetrics and Gynecology (4) recommend drug therapy for postmenopausal women who lack additional fracture risk factors at a T-score of 2.0 or less and for those with 1 or more additional fracture risk factors at a T-score of 1.5 or less. Miller and colleagues (9) suggested that a treatment T-score threshold of 1.8 may be reasonable for postmenopausal women even without additional fracture risk factors. As public health policy, however, such intervention makes sense only if the costs and risks of treatment are outweighed by a reduction in the ultimate cost and disability attributable to osteoporotic fractures. This is particularly relevant for osteopenia, since many postmenopausal women have low bone mass (10), and the societal direct medical cost of pharmacologic therapy for women with osteopenia would therefore be enormous. The cost-effectiveness of drug therapy may be particularly problematic in younger, early postmenopausal women, who, even if osteopenic, are not expected to be at high risk for fracture (11). We aimed to estimate the lifetime health benefits and costs of alendronate therapy to prevent fracture in postmenopausal women with osteopenia. Our analysis used the societal perspective. Methods We constructed a Markov cost-utility model that contained 8 health states and compared 5 years of treatment with alendronate (1 of the most commonly prescribed antiresorptive agents) with no drug therapy for women 55 to 75 years of age with varying levels of BMD T-scores (1.5 to 2.4). The health states we used were no fracture, post-distal forearm fracture, post-clinical vertebral fracture (that is, clinically evident at onset), post-radiographic vertebral fracture (that is, not clinically evident at onset), post-hip fracture, post-other fractures (that is, fracture of the proximal forearm, humerus, scapula, clavicle, sternum, ribs, pelvis, distal femur, patella, tibia, or proximal fibula), post-hip and vertebral fracture, and death. Women in the no fracture state can develop a distal forearm, hip, clinical vertebral, radiographic vertebral, or other fracture, at which time transition to that post-fracture state occurs. We assigned the direct and indirect costs of that fracture as transition costs. We modeled the disutility associated with these fractures as a lower value of a quality-adjusted life-year (QALY) associated with that fracture state. We assigned long-term care costs beyond the first year after hip fracture as a cost per year in the post-hip and vertebral fracture or post-hip fracture state. Individuals are eligible (at risk) to move to a different state once every 6 months. We assumed a discount rate of 3% for both costs and health benefits and a drug adherence rate of 100%. For the base-case analyses, we ran the model with 9 different combinations of starting age (55, 65, and 75 years) and femoral neck T-score (1.5, 2.0, and 2.4) until age 105 years, using Monte Carlo simulations with 40000 trials each, by using Data Pro HealthCare software (TreeAge Software, Inc., Williamstown, Massachusetts). Probabilities of Fractures We developed the risks for each type of fracture as a function of age from comprehensive population-based, age-specific data for women from the Rochester Epidemiology Project (12). This database captures almost all health care utilization within Olmsted County, Minnesota (13). Since an estimated 35% of all vertebral fractures in Olmsted County were clinically evident at onset (14), we set the incidence rate of radiographic (but clinically inapparent) vertebral fracture at 1.86 times that of clinical vertebral fracture. The rates for other fractures are the sum of incidences of the specific fracture types (12). We plotted fracture rates against the midpoint of each associated age range, and we determined a best-fitting power curve for each fracture as a continuous function of age. We adjusted each fracture risk function for bone density by using the method of De Laet and colleagues (15, 16), and we programmed the risk functions to change with increasing age and decreasing BMD for each stage of the Markov model. Relative Risk for Fracture during Drug Therapy The Fracture Intervention Trial (FIT) (17) of alendronate versus placebo enrolled many participants with osteopenia (femoral neck T-score, 1.6 to 2.5). On the basis of the clinical fracture group of FIT (women without any baseline radiographic vertebral fractures), we assumed relative risks for incident vertebral fractures of 0.54 and 0.82 for those with femoral neck T-scores of 2.0 to 2.4 and 1.5, respectively, and a relative risk for nonvertebral fractures of 1.0 for all osteopenic women receiving alendronate for the base-case analysis (17). We further assumed a linear, gradual offset of fracture reduction benefit over the subsequent 5 years after treatment as recommended by Tosteson and colleagues (18). Mortality We constructed an age-specific background mortality risk function from U.S. vital statistics for 2001 to model the risk for death for each cycle of the model (19). We estimated the mortality associated with acute hip fracture to be 1.375 times the base rate (20). Since the excess mortality associated with vertebral fracture may be attributable to preexisting comorbid conditions and not the fracture (21, 22), we assumed that excess mortality was not directly attributable to vertebral fractures or other nonhip fractures. Costs We assumed the yearly cost of alendronate (Table 1) to be the average U.S. wholesale price for 2001 ($842) (32) and that side effects from alendronate would generate only trivial direct medical costs. Table 1 also shows the direct medical costs associated with acute fractures (28), long-term care costs during the first and subsequent years after hip fracture (averaged for all patients with hip fracture) (29), and indirect costs of fractures (estimated in Meerding and colleagues' study [30]). The costs for 1 annual level-3 follow-up physician visit during alendronate therapy and for bone densitometry 2 years after alendronate therapy began (Table 1) are the median 2001 U.S. Medicare reimbursement rates for these services (31). Table 1. Model Parameters QALYs Associated with Each Health State For 1 year in the no fracture state, we used a QALY value estimated with the EuroQoL questionnaire (EQ-5D) in representative samples of the British and Swedish populations age 60 to 69 years (23, 24, 33). We derived the QALY values for the first and subsequent years with incident hip, distal forearm, clinical vertebral, and other fractures relative to the age-matched population from direct prospective estimates of Kanis and colleagues (25) (Table 1), and we derived the QALY values for the post-hip and vertebral fracture state from Tosteson and colleagues' study (27). We assumed radiographic (clinically inapparent) vertebral fractures to have a disutility of 0.08 (26), but only for 6 years after their occurrence since such fractures occurring more than 4 to 8 years ago do not seem to be associated with increased pain or limited activity (34, 35). Sensitivity Analyses We performed secondary analyses, assuming a gradual, linear 10-year offset of fracture reduction benefit after 5 years of drug therapy, combined with analyses modeling increased BMD-adjusted fracture risk because of additional risk factors to estimate how much additional risk is required for alendronate therapy to cost less than $50000 per QALY gained. We performed univariate sensitivity analyses, varying drug costs, discount rates, fracture rates, fracture costs, and the disutility from fractures and assuming preventable excess mortality for the first year after a clinical vertebral fracture. We performed 2-way sensitivity analyses, varying the relative risks for vertebral and nonvertebral fractures during alendronate therapy from 0.8 to 0.4 and from 1.0 to 0.6, respectively. Finally, we performed sensitivity analyses to model the possible effects of nonadherence to therapy, with alendronate used appropriately for only 2 years (with a 6-month offset of benefit) or used inappropriately for 2 years such that fracture reduction benefit is reduced by one third. We also performed probabilistic sensitivity analyses by using 2-stage Monte Carlo simulations (36, 37). The Appendix provides further details on the model str