Osteoporosis International

, Volume 16, Issue 12, pp 1545–1557

Fracture Reduction Affects Medicare Economics (FRAME): Impact of increased osteoporosis diagnosis and treatment


    • Public PolicyProcter & Gamble Pharmaceuticals
  • K. G. Saag
    • Department of MedicineUniversity of Alabama at Birmingham
  • R. T. Burge
    • Procter & Gamble Pharmaceuticals8700 Mason-Montgomery Rd
    • Ohio State University
  • M. Pisu
    • Center for Outcomes and Effectiveness Research and Education (COERE)University of Alabama at Birmingham
  • N. Goel
    • Procter & Gamble Pharmaceuticals8700 Mason-Montgomery Rd
    • University of Texas Medical Branch
Original Article

DOI: 10.1007/s00198-005-1869-5

Cite this article as:
King, A.B., Saag, K.G., Burge, R.T. et al. Osteoporos Int (2005) 16: 1545. doi:10.1007/s00198-005-1869-5


Osteoporosis is a common, debilitating disease affecting US Medicare beneficiaries, yet diagnosis and treatment lag behind medical advances. We estimated the cost of fractures to the Medicare program and the impact of increasing osteoporosis diagnosis and treatment. A Markov model was used to predict fracture incidence and costs in postmenopausal women aged 65 years and older, over 3 years (2001–2003). Only 1.80 million women were estimated to receive a Medicare-reimbursed bone mineral density (BMD) test in 2001. We evaluated the budget impact of testing an additional 1 million women from Medicare and patient perspectives. These women were stratified into high-risk (osteoporotic with prevalent vertebral fracture) and moderate-risk (without prevalent vertebral fracture) groups. During 2001–2003, an estimated 2.39 million fractures occurred among the 5.11 million women aged 65+ with osteoporosis, at a cost to Medicare of $12.96 billion. We projected that BMD testing of an additional 1 million women in 2001 would result in treatment of 440,000 new patients with a bone-specific medication, preventing over 35,000 fractures over the 3 years. The decrease in fractures would produce a net discounted savings to the Medicare budget of $77.86 million. Medicare’s hospital inpatient cost would decrease by $115.41 million and long-term care cost by $43.51 million, more than offsetting incremental outpatient cost of $81.07 million. Patients would benefit from fracture avoidance, but their out-of-pocket medical costs would increase by $63.49 million during 2001–2003, or $1,771 per fracture avoided. Sensitivity analyses showed that savings to the Medicare program varied in proportion to the unit cost of fractures, fracture risk of the populations tested, treatment rate, and adherence to therapy. Increased osteoporosis diagnosis may produce savings for the Medicare program if interventions are targeted to women at elevated risk for fracture and may be budget neutral if all older women are screened.


Bone densitometryEconomicsFracturesMedicareOsteoporosisTreatment


Osteoporosis is one of the most common and debilitating diseases affecting US Medicare beneficiaries, yet rates of diagnosis and treatment have lagged behind advances in diagnostic tests and availability of effective therapies [15]. In fact, fracture reduction was among 15% of the Healthy People 2000 objectives where the USA lost ground. From 1988 to 1998, the fracture rate per 100,000 increased from 714 to 863 in adults aged 65 and over, whereas the target was 607 [6]. Although Medicare coverage of bone density tests expanded in mid-1998 [7], Medicare paid for only 1.75 million bone density tests in 2000 [8], when over 20 million US women were aged 65 years or older [9].

In recent years, the US Congress considered legislation to further expand Medicare coverage of bone density tests, and the Centers for Medicare and Medicaid Services (CMS) conducted research to guide beneficiary education. These initiatives were hindered by lack of consensus on whom to screen and lack of data on the economic impact of osteoporosis on the Medicare program. The first issue was addressed in 2002 by the US Preventive Services Task Force, which recommended routine osteoporosis screening of women aged 65 years and older, as well as women aged 60 years and older at increased risk for osteoporotic fracture [10]. We conducted the FRAME (Fracture Reduction Affects Medicare Economics) study to address the hypothetical impact of further osteoporosis screening and treatment on the Medicare budget.

Materials and methods

A computer simulation model was used to predict the effects of osteoporosis testing and treatment on fracture incidence and associated costs in postmenopausal women aged 65 years and older in the Medicare program, over 3 years (2001–2003). We then evaluated the budget impact of testing an additional 1 million women from the Medicare and patient perspectives, and from the Medicare and patient’s combined perspective (Medicare+Patient). The study focused on postmenopausal women because they represent over 80% of US seniors with osteoporosis or osteopenia [11] as defined by the World Health Organization. Medicare payment and cost-sharing policies were assumed to remain constant.

For osteoporotic women, fracture incidence and outcomes were predicted using a Markov model (Fig. 1) that followed the population for 3 years. Tosteson and colleagues developed the underlying cohort-based model [12]. Adaptation of the model for population studies and the general methodology for determining epidemiology and cost inputs for US applications have been described previously [13].
Fig. 1

Movements between outcome states in Markov model of osteoporosis [12]

The effect of treatment on outcomes was based on published randomized controlled trials (RCTs) showing anti-fracture efficacy in osteoporotic women. For osteopenic women, published RCTs have demonstrated beneficial impact of medications on bone mass but not on fracture incidence [1416]. Therefore, this study conservatively assumed that incremental testing and treatment of osteopenic women would generate costs without affecting fracture outcomes.

Data sources: base case

To model policy and clinical practice in the base case, we undertook a descriptive analysis of disease prevalence, testing, and treatment in women aged 65 years and older in the Medicare program in 2001. These model inputs and data sources [9, 1732] are described in Table 1 and Appendices A and B. The model accounted for population growth during 2001–2003 but held practice patterns constant over the 3-year projection.
Table 1

Model inputs: epidemiology, BMD testing, and treatment in US women aged 65 years and older (Vert vertebral, Fx fracture, BPs bisphosphonates, SOF Study of Osteoporotic Fractures)



Source description and comments


Population, aged 65–90 years








Number and proportion (%) with osteoporosis; osteopenia; normal bone mass

5,113,764 (25.0); 9,818,425 (48.0);

[17] NHANES III femoral neck BMD

5,522,856 (27.0)

Proportion of osteoporotic women with prevalent vert fx (high risk); without (moderate risk) (%)

38.5; 61.5

[18]Rochester, Minnesota population

Hip fx incidence rates


[19] 1997 Nationwide Inpatient Sample of over 1,000 community hospitals in 22 states across all payers; sampling weights used to calculate national estimates

Appendix A

Vert and wrist/forearm fx incidence rates


[20,21] Rochester, Minnesota, population; linear smoothing to convert 5-year age bands to age-specific rates

Appendix A

Proportion of hospital fx cases attributable to osteoporosis, by age group

65–84, 85+ years

[22] Attribution probabilities for white women, based on Delphi panel

  Hip fx

0.90, 0.95

  Vert fx

0.90, 0.95

  Wrist/forearm fx

0.75, 0.80

Relative risk of fx by fx site and risk group, by 5-year age band


Proportion with low BMD: [17]; relative risk: [23] and SOF data, personal communication, D.M. Black, July 26, 2002

Appendix B

General mortality, rate per 1,000 women, by age


[24] Social Security Administration

Excess mortality following hip fx, rate per 1,000 women, by age


[25], reduced by 50% for comorbidities that contribute to mortality [12]

BMD testing and treatment

Medicare BMD tests, 2000 (No.)


[8] Center for Medicare and Medicaid Services, Part B extract and summary system; assuming all tests in women aged 65+years, each test represents a unique individual, and 10% testing growth from 2000 to 2001

Proportion of BMD tests in women (%)


[26] MedSTAT, MarketScan data, 1998–2000

Proportion of tested women who have osteoporosis; osteopenia (%)

50.0; 50.0

Kaiser Permanente population subset aged 65+ years, unpublished data from study reported in [27]; personal communication, Alice Pressman, February 13, 2002

Proportion of treated women who have osteoporosis; osteopenia (%)

69.4; 30.6

Treatment with bone-specific agents

See below

[26] MedSTAT, MarketScan drug claims for women aged 65+ years with prescription drug benefit and continuous enrollment in Medicare supplemental or retiree plan

Proportion of tested women who are treated with bone agent (%)


[26] Enrollment 1/99–7/00; treatment with BP, raloxifene, or calcitonin during 1/98–12/00

Women who are treated with bone agent without a BMD test in preceding 12 months (%): Proportion of total women; of women with osteoporosis or osteopenia; of women with osteoporosis

11.5; 15.9; 10.9

[26] Proportion of total women: drug claims 1/99–7/00; proportions for subpopulations adjusted for all tests occurring in women with osteoporosis or osteopenia, and an equal number of tests in osteoporotic and osteopenic women

Therapy discontinuation at 3 months, 1 year, 2 years


[26] Enrollment 1/97–12/00; drug claims 1/99–12/00

Efficacy (relative risk reduction) by fx site and risk group (moderate, high)


[14, 15, 16, 28, 29, 30,31] Randomized controlled clinical trials

Therapy market share, 2001 (%)

[32] IMS Health, national prescription data for women aged 65+ years with osteoporosis diagnosis and prescription for a bone-regulating therapy

BPs, calcitonin, raloxifene

70.2, 17.1, 12.7


Over 5 million US women aged 65 years and older were estimated to have osteoporosis in 2001 and nearly 10 million to have osteopenia (Table 1). To project fracture incidence among the osteoporotic women, the Markov analysis utilized age-specific fracture incidence rates for women (Appendix 1). Hip fracture incidence was estimated from the Nationwide Inpatient Sample (NIS) because virtually 100% of hip fractures lead to a hospital admission. The NIS rates were adjusted downward to reflect the proportion of hospital fracture cases that were attributable to osteoporosis [33], using published attribution rates for white women [22]. Previous research has shown whites to account for over 93% of osteoporotic fracture costs in the USA [34].

We stratified the osteoporotic women by severity of disease into two groups: (a) high-risk individuals with osteoporosis and a prevalent vertebral fracture [35] and (b) moderate-risk individuals with osteoporosis and no vertebral fracture. To predict the number of fractures within the study population of osteoporotic women, the fracture incidence rates for women aged 65 years and older were adjusted for the relative risk of fracture in women with low bone mineral density (BMD), with or without a prevalent vertebral fracture [23] (D. Black, personal communication, July 26, 2002).

Osteoporosis testing and treatment

In 2001, osteoporosis diagnosis and treatment were uncommon among older women; only 1.80 million women (8.8%) aged 65 years and older were estimated to receive a Medicare-reimbursed BMD test (Fig. 2) [8]. For our base case, we assumed that this small number of women obtained a BMD test because they had at least one risk factor for fracture and, in accordance with guidelines of the National Osteoporosis Foundation [36], a BMD test could influence a treatment decision. In the absence of a better measure of fracture risk, we assumed that in 2001 all women tested through the Medicare program would have low BMD (osteoporosis or osteopenia), and that BMD status would affect the likelihood of treatment. Assumptions about the characteristics of patients who received BMD testing were further examined in sensitivity analyses discussed below.
Fig. 2

Medicare-reimbursed BMD tests in men and women, 1997–2001 [8]

We assumed that the tested women were split equally among osteoporotic and osteopenic, based on BMD test results in women aged 65 years and older in a managed care plan [27]. The overall proportion treated following a BMD test, 44%, was based on data from Medicare supplemental and retiree plans [26], with greater treatment of osteoporotic women (61%) than osteopenic women (27%) [27]. Thus, we estimated that 17.5% of osteoporotic women and 9.2% of osteopenic women were tested in 2001, and that 71.6% of osteoporotic women (Fig. 3) and 86.1% of osteopenic women received neither treatment nor a Medicare-reimbursed BMD test. Among those tested, half (those with osteopenia) were assumed to incur the patient share of BMD testing and treatment costs and to have no measurable beneficial effect of therapy on fracture incidence.
Fig. 3

Estimated BMD testing and treatment (TX) in 5.1 million osteoporotic women aged 65+ years, 2001. Data sources: CMS BESS [8], MEDSTAT MarketScan [26], Kaiser Permanente [27] (personal communication)

For osteoporotic women, the Markov model simulated the effects of three treatments on fracture incidence. We included only FDA-approved bone-specific osteoporosis medications in the analysis, specifically the bisphosphonates (BPs) risedronate (Procter & Gamble Pharmaceuticals) and alendronate (Merck) as a class, raloxifene (Eli Lilly), and nasal calcitonin (Novartis Pharmaceuticals). All medication inputs were weighted by market share of each product in 2001. We conservatively assumed zero efficacy when data were unavailable or not statistically significant (at p <0.05) for a particular medication and fracture site. We also assumed no residual efficacy following discontinuation of medication and no efficacy at all for women discontinuing during the first 3 months. Discontinuation rates were 58% (BPs), 63% (raloxifene), and 92% (calcitonin) at 2 years [26].

Cost of osteoporosis testing, treatment, and outcomes

Table 2 summarizes data sources [34, 3748] and assumptions for cost of fractures, BMD tests, and medications. To estimate cost of fractures to Medicare, national inpatient hospital data were combined with published data on outpatient and long-term care (LTC), which included care in nursing facilities and disability/dependence. Unit costs were determined for the age groups 65–74, 75–84, and 85+. For vertebral and wrist/forearm fractures, the mean costs for health care services were weighted by the low proportion of patients who receive those services [13]. All costs were converted into 2001 terms using the Inpatient Hospital Care or Medical Care services component of the urban Consumer Price Index [49]. A 5% discount rate was applied to costs and fractures over the 3-year study period.
Table 2

Model inputs: Costs of fractures, BMD tests, and medications to Medicare in 2001 (BPs bisphosphonates, DRG diagnosis-related groups, Fx fracture, LTC long term care, vert vertebral, Rehab rehabilitation, SNF skilled nursing facility, LOS length of stay, AWP average wholesale price)



Source description and comments

Hospital inpatient unit cost ($), by age group

65–74, 75–84, 85+ years

See below for components of hospital unit cost

  Hip fx

16,627, 16,208, 15,881

  Vert fx

1,031, 926, 806

  Wrist/forearm fx

617, 617, 617

Inpatient facility

  Mean charge, by age group


[19] Primary fx diagnosis, high cost outlier trim

  Medicare cost to charge ratio



Inpatient physician cost ($)

  Hip fx



  Vert fx


[39] and Medicare fee schedule for CPT codes 99222, 99235, and 99239 [40]

Additional hip fx hospital costs ($)

  Emergency room services



  Readmission to acute care hospital


[41] 8.0% readmitted to acute-care hospital

  Admission to rehab or short-stay hospital


[41] 12.0% admitted within 1 year

Inpatient adjustment factor

[13], Cost per vert and wrist/forearm fx adjusted for low proportion of patients receiving hospital care

  Vert fx


  Wrist/forearm fx


Outpatient unit cost ($)

All age groups

  Hip fx



  Vert fx



  Wrist/forearm fx


0.508 × hospital inpatient cost [34]

Physician visits (number), hip and vert fx

  1-year post-hospital, hip fx; vert fx

8; 2.3

Hip [41]; Vert [39]

  Without hospitalization, vert fx



LTC unit cost ($), by age group

65–74, 75–84, 85+ years

See below for components

  Hip fx

4,154, 4,801, 5,607

[45] MedPar SNF 2000

  Vert fx

234, 309, 399


  Wrist/forearm fx

138, 138, 138

0.224 × hospital inpatient cost [34]

Disability/dependence ($), hip; vert

1221; 75


Hospital discharges to SNF (%), by age group

65–74, 75–84, 85+ years

Primary fx diagnosis, high-cost outlier trims

  Hip fx

42.9, 48.7, 56.6


    Following discharge to home

6.0, 6.0, 6.0


  Vert fx

28.1, 40.3, 51.7


Medicare reimbursement for SNF admission following hip or vert fx, by age group


2000 Medicare Expanded MedPAR File, SNF, for DRGs 209, 210, 211, 236, 239 [45]

LOS, ratio full to partial year



Private-pay per diem cost ($/d)


[43,44]: range 121–168

BMD test cost ($)


[8] CMS BESS 2000 weighted average of Medicare allowed charges HCPCS 76075, 76076, 7697, G0130, 78350, 78351

Medicare; patient share (%)

75; 25

Medication cost

Third-party payers

85% AWP


Patients paying co-pay


Standard co-pay

Patients paying cash

100% AWP


Cash proportion of prescriptions

BPs, calcitonin, raloxifene (%)

14.1, 15.3, 11.6


Serious side effects

BPs ($)


[47] BPs, per treated patient, first year of therapy

Raloxifene ($)

1,527/venous thromboembolism

[48] Raloxifene; excess event rate 1.72/1000 patient years (Eli Lilly, data on file, personal communication, L. Plouffe and A. Herrold, October 1, 2001)

In accordance with Medicare cost-sharing requirements, Medicare beneficiaries were assigned the cost of the 20% coinsurance for Part B services, Part A deductible for inpatient hospital stays ($700), and Part A per diem co-payments for stays longer than 20 days in a skilled nursing facility (SNF). After the 100-day limit for Medicare SNF coverage, 100% of nursing home costs were assigned to patients who lacked supplemental insurance. Our analysis assumed a 2:1 ratio of full-year to partial-year stays in nursing facilities [41].

BMD costs were based on a weighted average of Medicare’s allowed charges, which ranged from $11.59 for single energy X-ray absorptiometry (GO130) in the outpatient hospital setting to $151.95 for axial dual energy X-ray absorptiometry (DXA, 76075) in a skilled nursing facility. Axial DXA tests constituted 90% of the total number of allowed services, with an average allowed charge of $94.44. Elimination of low-cost tests (average charge <$30) increased the weighted average BMD test cost by 28%, which is similar to the 25% increase tested in our sensitivity analysis.

The costs of BMD tests, medications and side effects in osteopenic women were calculated from costs predicted for osteoporotic women, based on the relative testing and treatment rates for these two subpopulations. For women with normal bone density, the cost of BMD tests was evaluated in sensitivity analyses.

Impact of additional testing: primary analysis

We evaluated the impact of testing an additional 1 million women in year one (2001), approximately a 56% testing increase. As in the base case, we modeled clinical practice in 2001 and assumed that 500,000 new tests were for women who would be diagnosed with osteoporosis and 500,000 were for women who would be diagnosed with osteopenia. Thus, the policy change simulated incremental testing of approximately 10% of the osteoporotic population and 5% of the osteopenic population aged 65 years and older.

Impact of additional testing: sensitivity analyses

One-way sensitivity analyses evaluated population (disease prevalence, moderate/high risk proportion, osteoporosis/osteopenia treatment proportion), therapy (proportion of women with BMD tests who obtain therapy, proportion of women who were on therapy but did not have a BMD test, therapy discontinuation rates, cash payer proportion of prescription drugs, adverse event costs), and economic parameters (unit costs of fractures, unit cost of BMD test, discount rate). Multi-way analyses were conducted within categories and using all variables across categories. Best- and worst-case scenarios were constructed to yield the highest and lowest savings for Medicare, respectively, under 2001 practices.

In addition, we conducted one-way analyses to predict the impact of liberalizing Medicare BMD coverage policy to allow routine screening of all estrogen-deficient women. We assumed either that the results of the BMD testing would reflect the population distribution of bone density (27% normal, 48% osteopenic, 25% osteoporotic), or that half of women with low BMD results would be osteopenic and half osteoporotic (27% normal, 36.5% osteopenic, and 36.5% osteoporotic).


Base case

During 2001–2003, 2.39 million fractures were estimated to occur among the 5.11 million women with osteoporosis who were aged 65 years and older (Table 3), at a cost to Medicare of $12.96 billion. Hospitalization constituted 68% of total Medicare costs and long-term care 25%, whereas BMD testing and other outpatient costs were only 7%. For patients, long-term care constituted 65% of costs, whereas hospitalization was only 19% and outpatient care 16%.
Table 3

Primary analysis: Impact of additional BMD testing and treatment on fractures in osteoporotic women, by risk group, 2001–2003. (Moderate risk group includes osteoporotic women without a prevalent fracture. High risk group includes osteoporotic women with a prevalent fracture. Numbers in parentheses are negative)

Fracture risk group




Estimated population, 2001




Predicted fractures, 2001–2003

   Base case




   1 million additional BMD tests




   Net change in fractures




Impact of additional testing: primary analysis

This study predicts that bone density testing of an additional 1 million women aged 65 years and older would result in treatment of an additional 440,000 women with a bone-specific medication, preventing over 35,000 fractures during 2001–2003 (Table 3). The decrease in fractures would produce a net discounted savings to the Medicare budget of $77.86 million over the 3-year period (Table 4). Medicare’s hospital inpatient costs would drop by $115.41 million and long-term care costs by $43.51 million, more than offsetting incremental outpatient cost of $81.07 million (Fig. 4). The increased outpatient cost stems from incremental expenditure of $67.32 million for BMD tests and $19.87 million for adverse events associated with medications, offset by $6.13 million in savings from fewer office visits and other outpatient care. Although women with a prevalent vertebral fracture represent only 38.5% of those with osteoporosis, new BMD tests and treatment in this high-risk group account for over half of the fracture reduction (Table 3) and two-thirds of the Medicare savings, or $51.94 million.
Table 4

Primary analysis: Impact of additional BMD testing and treatment on medical costs in women aged 65 years and older, by cost perspective and testing scenario, 2001–2003

Costs (USD) by perspective, 2001–2003





Base case




1 million additional BMD tests




Net discounted cost




Net discounted cost per fracture avoided




Primary analysis assumes additional 1 million tests include 500,000 in osteoporotic women and 500,000 in osteopenic women. Costs reported in 2001 dollars with 5% discount rate. Numbers in parentheses are negative and represent savings. Number of fractures avoided was predicted to be 35,858 (Table 3)

Fig. 4

Net cost of 1 million additional Medicare BMD tests in women aged 65+ years, 50% with osteoporosis and 50% with osteopenia, 2001–2003. Costs reported in 2001 dollars with 5% discount rate

Patients would benefit from fracture avoidance, but their out-of-pocket medical costs would increase by $63.49 million. This represents an incremental cost of $1,771 per fracture avoided. Over the 3 years, added spending of $115.41 million for osteoporosis medications would outweigh patients’ savings of $4.62 million from reduced out-of-pocket costs for other outpatient care, $12.40 million for hospital inpatient care, and $42.44 million for LTC (Fig. 4). Patients would also pay an incremental $2.22 million for BMD tests and $4.96 million for medication-related adverse events.

Impact of additional testing: sensitivity analyses

Medicare perspective

Sensitivity analyses support the conclusion that an increase of 1 million BMD tests would lead to budget savings or budget neutrality. Budget decreases (savings) were achieved in all scenarios except for the worst-case scenario across all ten parameters, where an increase of $54.5 million was estimated. In the other best- and worst-case scenarios, budget decreases to Medicare ranged from $237.6 million to $1.5 million (Table 5). Among population and therapy parameters, the results were most sensitive to the overall treatment rate following a BMD test, the relative treatment rates for osteoporotic and osteopenic women, and the therapy discontinuation rate. Among economic parameters, Medicare savings were most sensitive to changes in the unit cost of fractures. Medicare savings exceeded incremental costs to patients, except under scenarios where treatment of osteoporotic women decreased, treatment rates equalized for osteoporotic and osteopenic women, or the proportion of treated women paying cash for medications increased.
Table 5

Sensitivity analysis: effects of varying model inputs on the Medicare budget impact of 1 million additional BMD tests. (Numbers in parentheses are negative)

Model inputs

Impact of additional BMD testing on Medicare budget, 2001–2003

Variable (estimate in primary analysis)

Estimate in sensitivity analysis

Difference from base case* ($)

Primary analysis†



Population variables

Prevalence of osteoporosis and osteopenia (25% and 48%)

10% decrease


10% increase


Proportion of osteoporotic women with a prevalent vertebral fracture (61.5%)





Proportion of treated women who have osteoporosis; osteopenia (69.4%; 30.6%)

50%; 50%


75%; 25%



Best case


Worst case


Therapy variables

Therapy discontinuation

10% decrease


10% increase


Adverse event cost

25% decrease


25% increase


Overall proportion treated with bone agent following BMD test (44%)







Proportion with osteoporosis therapy but no BMD test (11.5%)






Best case


Worst case


Economic variables

Unit cost of fractures

25% decrease


25% increase


Cost of bone density test

25% decrease


25% increase


Discount rate (5%)






Best case


Worst case


Multi-way across all variable categories

All variables above

Best case


Worst case


New policy scenario: screening women with normal BMD

Proportion of tested women who have osteoporosis; osteopenia; normal BMD (50%; 50%; 0%)

25%; 48%; 27%


36.5%; 36.5%; 27%


*Under the base case, the estimated cost of fractures was $12,958,209,568 in 2001–2003

The primary analysis predicted that testing an additional 1 million women would result in a cost to Medicare of $12,880,351,755 in 2001–2003 and a difference from the base case of ($77,857,814)

**Includes all variables within category

A change in policy and medical practice that substantially increased Medicare screening of women with normal BMD would reduce savings to the Medicare program. If a woman’s BMD status is assumed to have no impact on the likelihood of receiving a Medicare-reimbursed BMD test, the Medicare program would realize a nominal savings of $0.70 million over the 3 years modeled.

Patient perspective

The simulation models were most sensitive to the proportion of patients who pay cash for medications and the treatment rate following a BMD test. A 25% increase in the proportion of cash payers resulted in a net patient cost of $88.92 million, relative to the base case, whereas, a 25% decrease in cash payment produced a net patient cost of $38.05 million. Changing the overall treatment rate from 44% to 22%, 33%, or 55% resulted in net patient cost of $32.5 million, $48.00 million or $78.98 million, respectively.


US guidelines recommend routine screening for osteoporosis in women aged 65 years and older [10] and consideration of treatment for those with bone density scores of −2.0 or lower, or −1.5 with at least one risk factor for fracture [36]. This study determined the monetary value to the US Medicare program and patients of helping to close the gap between these guidelines and clinical practice. The study predicts an annual fracture rate of 1,556 per 10,000 women aged 65 years and older with osteoporosis, indicating the need to accelerate interventions to identify women at elevated risk for fracture.

Our research demonstrates that increased diagnosis and treatment of osteoporosis among women in the Medicare program would produce Medicare savings under a range of conditions. The magnitude of Medicare savings would vary in proportion to the unit cost of fractures, fracture risk of the populations tested, the treatment rate following BMD testing, and adherence to therapy. Our base case assumed that in 2001 Medicare reimbursed a negligible number of BMD tests in women with normal bone density. At that time, BMD testing and disease awareness were low among senior women overall [5], and 70% did not know that Medicare paid for the test [50]. The 1996 guidelines of the US Preventive Services Task Force had found insufficient evidence to recommend routine screening in postmenopausal women [51], and Medicare covered the BMD test only for diagnosing, treating, or monitoring qualified beneficiaries [7]. For a woman to qualify for Medicare BMD coverage on the basis that she was estrogen-deficient, federal regulation required documentation of “clinical risk for osteoporosis, based on her medical history and other findings” [7]. In recent years, however, there has been growing support for routine screening of postmenopausal women aged 65 years and older [10, 52]. We evaluated the impact of a more liberal Medicare BMD coverage policy in a sensitivity analysis that assumed fracture risk had no bearing on a woman’s likelihood of obtaining a Medicare BMD test. This scenario produced a small savings, which might be considered budget neutral in view of Medicare’s total expenditures of over $200 billion in 2001.

Our primary analysis showed the potential benefit of targeting Medicare interventions toward the subgroup of older women who have already experienced a vertebral fracture. Even among these high-risk patients, the gap between real-world practice and quality care persists [24, 53]. In a managed care population studied in 1998–2001, fewer than 5% of women with new fractures had a BMD test either in the year before or the 6 months following the fracture, and fewer than half (45.3%) received medication for osteoporosis [5]. A new HEDIS (Health Plan Employer Data and Information Set) measure for Medicare health plans in 2004, “Osteoporosis management in women who had a fracture,” will quantify the number of women who receive either a BMD test or prescribed medication for osteoporosis during the 6 months following a fracture [54]. Although this measure will not document whether appropriate care follows a BMD test, it should create incentives for secondary prevention.

Several large epidemiologic studies suggest that other risk factors should be used in conjunction with BMD to identify high-risk patients [5557]. Data from the Study of Osteoporotic Fractures indicate that although low BMD predicts fractures, fewer than half of fractures may be caused by, or attributable to, low BMD in older women [57]. These data are consistent with the US National Institutes of Health definition of osteoporosis as compromised bone strength predisposing to fracture [33]. There is increasing evidence that factors such as bone architecture [58] contribute to bone strength. In our study, BMD status was used as a proxy for fracture risk and determined both the likelihood of treatment and the efficacy of treatment. Therapy was assumed to prevent fractures only in women with osteoporosis defined as a T -score less than −2.5. In effect, our assumptions modeled the incremental testing of 1 million women who were at risk for fracture, assuming that treatment would be initiated in 440,000 and that 305,000 of these women might show fracture response to therapy according to efficacy rates shown in clinical trials, adjusted for the duration of therapy (8–42% remaining on medication at 2 years).

Data limitations presented numerous challenges to the modeling of osteoporosis practices and interventions. Excess mortality following hip fracture was based on data from a European population, and incidence rates for vertebral and wrist/forearm fractures were derived from a Minnesota population. In addition, we relied on data from managed care and employer-sponsored health plans [26,27] to estimate BMD testing, osteoporosis treatment, and therapy discontinuation in older women. Although these populations were not representative of all US women aged 65 years and older, they may reasonably approximate the medical care behavior of women who received Medicare BMD tests in 2001, and they served as a data-driven starting point for our base case. Women with supplemental coverage and hence low out-of-pocket costs for pharmacy and medical care would be more likely to obtain preventive services, such as BMD screening and medication, for an asymptomatic condition like osteoporosis [5960].

For patients, the net costs of increased testing and treatment were driven by assumptions affecting their out-of-pocket payments for medications. The primary analysis assumed patients paid for only 11.6% to 15.3% of the prescribed osteoporosis therapies. US prescription data indicated that the vast majority of older women who received prescriptions for bone-specific medications had third-party coverage [32]. Our analysis may have overestimated costs for patients paying out-of-pocket for medicines, because they were assumed to pay an undiscounted average wholesale price (AWP). During the years modeled by this study, pharmacy discount cards proliferated. Several states implemented programs wherein Medicare beneficiaries could access Medicaid pharmacy discounts of 10% or more off AWP. In California and Florida alone, these discounts were available to 1.85 million men and women in the Medicare program who lacked drug coverage. Cards administered by pharmacy benefit managers had median discounts on brand-name products ranging from $2.09 to $20.95 for a 30-day supply [61] or an average of 14% off retail price [62]. In addition, low-income individuals without prescription coverage could obtain free or heavily discounted medicines through state or manufacturer patient assistance programs.

Our modeling may also have overestimated patients’ out-of-pocket costs for hospital and outpatient care, because supplemental insurance coverage was taken into account only for medication and skilled nursing facility costs. Almost 88% of Medicare beneficiaries have supplemental insurance, including 24% with Medigap coverage [63]. All standardized Medigap plans cover both Part A (hospital) and Part B (outpatient) coinsurance, and in 1999, over 60% of purchasers of standard Medigap plans enrolled in plans that covered both Part A and Part B deductibles [64].

Overall, we believe this study presents a conservative estimate of the benefits to Medicare and its beneficiaries of preventing osteoporotic fractures for several reasons. The prevalence of osteoporosis, and hence fracture incidence, may have been underestimated by use of NHANES III data on femoral neck BMD, which may not reflect bone loss at other sites. In addition, the structure of the model limited fracture incidence to one per year. We also did not evaluate indirect or long-term costs of fracture. Braithwaite and colleagues estimated the lifetime attributable cost of hip fracture to be $81,300 (2001 dollars) in a cohort of 80-year-olds. LTC accounted for nearly three-fourths of this cost. Their study included indirect costs such as informal home care, which constituted 30% of the total fracture cost [65]. They predicted that 227 days in a nursing facility would be attributable to hip fracture, which is consistent with other reports truncating nursing home costs at 1 year [38].Because LTC services comprise 65% of Medicaid spending for dual eligibles [66], our analysis underestimated government savings from fracture prevention in that we did not quantify savings to the Medicaid program.

Conservative assumptions were also employed regarding therapeutic interventions. The cost of gastrointestinal side effects for both bisphosphonates, $10.54 per treated patient per month, were based on a single published study, and more recent data suggest a lower average monthly cost per patient of $2.52 for risedronate and $7.40–$7.50 for alendronate [67]. In addition, the analysis did not account for growth in treatment over time, which would produce additional savings. For example, the results of the Women’s Health Initiative (WHI) trial, announced in 2002, [68] may increase utilization of non-hormonal therapies for prevention and treatment of osteoporosis. Any increases in adherence to medications would also increase the cost effectiveness of diagnosis and treatment. This may be counterbalanced, however, by the fact that osteoporosis treatment rates in our analysis were based on data from populations with prescription benefits, which may have higher utilization and adherence than in the overall Medicare population in 2001–2003.

This study projects that testing approximately 10% of senior women with osteoporosis and 5% of those with osteopenia would save Medicare $77.86 million over 3 years. Our projection of savings is supported by an observational study, which estimated that increasing osteoporosis testing and treatment through disease management produced a savings of $10.3 million over 5 years in women aged 65 years and older enrolled in a single managed care plan [69]. It is noteworthy that in our study Medicare savings would exceed total incremental costs to patients.

Under current policy, patients and third-party payers bear the lion’s share of costs for osteoporosis medications, whereas, Medicare reaps the greatest financial benefit from fracture prevention. Lack of prescription coverage is a substantial barrier to access to osteoporosis therapies. By instituting outpatient coverage of prescription medications, under the voluntary Medicare Part D effective January 2006, the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 may shift a significant portion of osteoporosis medication costs from patients to Medicare.

The Medicare Modernization Act (MMA) also established a preventive physical exam for new Medicare enrollees, effective January 2005, that includes height measurement, counseling and referral for a BMD test. The implementing regulation requires the exam to include a medical and social history “with attention to modifiable risk factors” and advises that health professionals will be able to ask about risk factors such as height loss and previous fracture. The patient must also receive a brief written plan for appropriate screening and preventive services [70]. The preventive physical exam should help identify patients at elevated fracture risk, for whom BMD testing and follow-up are health priorities. In addition, the Medicare Part D benefit may improve access to therapies.

Our study estimates the impact of increased osteoporosis diagnosis and treatment during a period prior to major change in the Medicare program. Further research is needed to better quantify current medical care for individuals with osteoporosis, osteopenia, and fracture risk factors. This baseline would enable future evaluation of the changing cost and fracture burden for patients and the impact of disease management and quality-of-care initiatives on clinical practice. A study of the full societal cost of fractures would be valuable to capture the shifting roles of the fee-for-service Medicare program, Medicaid, and private health plans following implementation of the MMA.

Modeling suggests that policies that increase diagnosis of osteoporosis and appropriate treatment of women in the Medicare program may produce savings if interventions are targeted to women with risk factors for fracture and may break even if all older women are screened. Such policies may represent good value from the patient perspective, as well, when the Medicare program begins subsidizing outpatient prescription medications. Greater detection of osteoporosis and increased use of effective therapies appear to be both good medicine and sound economics.


The authors thank Karen Worley for consultation, Dan Worley and Eric Balda for programming, and Lorraine Rice for administrative assistance.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2005