Objective: To examine the life expectancy and cost-effectiveness of hormone replacement therapy (HRT) and raloxifene therapy in healthy 50-year-old postmenopausal women.

Methods: We performed a cost-effectiveness analysis using a Markov model, discounting the value of future costs and benefits to account for their time of occurrence.

Results: Both HRT and raloxifene therapy increase life expectancy and are cost-effective relative to no therapy for 50-year-old postmenopausal women. For women at average breast cancer and coronary heart disease risk, lifetime HRT increases quality-adjusted life expectancy more (1.75 versus 1.32 quality-adjusted life years) and costs less ($3802 versus $12,968) than lifetime raloxifene therapy. However, raloxifene is more cost-effective than HRT for women at average coronary risk who have a lifetime breast cancer risk of 15% or higher or who receive 10 years or less of postmenopausal therapy. Raloxifene is also the more cost-effective alternative if HRT reduces coronary heart disease risk by less than 20%.

Conclusions: Assuming the benefit of HRT in coronary heart disease prevention from observational studies, long-term HRT is the most cost-effective alternative for women at average breast cancer and coronary heart disease risk seeking to extend their quality-adjusted life expectancy after menopause. However, raloxifene is the more cost-effective alternative for women at average coronary risk with one or more major breast cancer risk factors (first-degree relative, prior breast biopsy, atypical hyperplasia or BRCA1/2 mutation). These results can help inform decisions about postmenopausal therapy until the results of large scale randomized trials of these therapies become available.

^aDepartment of Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA
^bDepartment of Gynecology and Obstetrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA
^cCenter for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA
^dUniversity of Pennsylvania Cancer Center, Philadelphia, Pennsylvania, USA
^eLeonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia, Pennsylvania, USA

Dr. Armstrong is supported by an American Cancer Society Clinical Research Training Grant 99-023-01 and Grant BC971623 from the Department of the Army Breast Cancer Research Program. Dr. Schwartz is supported by a National Cancer Institute Comprehensive Cancer Center Grant and Grant BC971623 from the Department of the Army Breast Cancer Research Program. Dr. Albert is supported by a grant from NIAMS (PO 1 AR 495584).

Henry Glick, PhD, provided invaluable assistance with regression models for coronary heart disease.

(Obstet Gynecol 2001:98:996-1003. © 2001 by The American College of Obstetricians and Gynecologists.)

Deciding about the use of hormone replacement therapy (HRT) or raloxifene after menopause is difficult. These therapies have multiple, often competing effects.^1-7 The most effective method of extending life expectancy depends upon an individual woman's risk for osteoporotic fracture, coronary heart disease, or breast cancer, and the relative efficacy of these therapies on reducing these events. Synthesizing this complex information is made particularly difficult by the large number of often conflicting studies and the need to extrapolate the efficacy of raloxifene on clinical outcomes from surrogate endpoints and the efficacy of HRT from observational studies.^5,6,8 Furthermore, differences in prescription drug costs of raloxifene therapy and HRT suggest that the short- and long-term economic costs of these therapies may vary substantially.

In this setting, decision analysis offers a systematic approach to evaluating the comparative clinical and cost-effectiveness of alternative therapies, including the impact of alternative assumptions on outcomes of interest. The objective of the present study was to examine the life expectancy and cost-effectiveness of HRT and raloxifene therapy to prevent the long-term complications of estrogen deficiency among healthy postmenopausal women.

Materials and Methods

Clinical and cost-effectiveness were estimated using a time-dependent Markov model that simulated the outcomes of HRT, raloxifene, or no therapy in hypothetical cohorts of 50-year-old healthy postmenopausal women. The simulation included the six major outcomes affected by raloxifene and HRT: coronary heart disease, vertebral fracture, hip fracture, thromboembolism, endometrial cancer, and breast cancer. Because data about the impact of HRT on colon cancer and Alzheimer's disease are preliminary and corresponding data are not available for raloxifene, these outcomes were not included in the simulation. Risks of developing each outcome were independent of prior outcomes. The simulation was run until all cohort members died or reached age 101.

The analysis compared three alternative regimens: HRT (0.625 mg of oral conjugated estrogen per day with cyclic progestin for 10-14 days per month in women with an intact uterus); raloxifene (60 mg per day); and no treatment. All women were assumed to be compliant with therapy. The base-case analysis examined use of continuous therapy from age 50 until death. Because some women take HRT or raloxifene for shorter time periods, therapy of 5- and 10-years duration after menopause at age 50 was examined in secondary analyses, with benefits of therapy assumed to continue only while therapy was used.

Simulation outcomes included life expectancy, quality-adjusted life years, and direct medical cost.⁹ Although the inclusion of all direct medical costs is consistent with a societal perspective, nonhealth effects, health effects on people other than the woman in question, indirect medical costs and nonmedical costs are not currently able to be measured adequately and were not included in the analysis.⁹ Costs and benefits were discounted at a 3% annual rate to account for their decreased value over time.⁹ The model was validated by comparing the life expectancy of a 50-year-old woman at average cardiac and breast cancer risk who selects no therapy from the simulation (31.68 years) to estimated life expectancy of a 50-year-old US woman from the National Center for Health Statistics (31.7 years).¹⁰

Transition probabilities for disease incidence, disease mortality, and the impact of alternative therapies on disease incidence are shown in Tables 1 and 2. For the base-case analysis, the probability of developing coronary heart disease was that of women with population levels of low-density lipoprotein, total cholesterol, systolic blood pressure, no history of diabetes, smoking, or left ventricular hypertrophy.^11,12 The effect of raloxifene on coronary heart disease was estimated from its impact on total cholesterol and high-density lipoprotein in the base-case analysis and its impact on low-density lipoprotein in sensitivity analyses.^12,13 The effect of HRT on coronary heart disease was taken from a large, prospective cohort study in the base-case and its impact on lipids in sensitivity analyses.^1,13 Consistent with the results of a recent randomized controlled trial of HRT in women with coronary heart disease (HERS), HRT was assumed not to affect mortality after a diagnosis of coronary heart disease.⁸ Estimated mortality after a diagnosis of coronary heart disease was adjusted for the recent substantial decrease in coronary heart disease case fatality among US women.^14,15

The relative risk of hip fracture in the base-case analysis was determined from the effect of raloxifene on bone mineral density, and the point estimate from the MORE study was examined in sensitivity analyses.^4,5 Vertebral fractures were assumed to affect costs and quality of life but not life expectancy. The effect of HRT on vertebral fracture was assumed to be equal to its effect on hip fractures. Although HRT was assumed not to increase the risk of endometrial cancer in the base-case analysis, increases in endometrial cancer risk were examined in sensitivity analyses. Mortality from other causes was obtained by subtracting mortality from the outcomes included in the model from all-cause mortality rates.¹⁰

Cost and utility model parameters are shown in Table 3. Direct medical costs included average wholesale medication acquisition costs for HRT and raloxifene (obtained from the Red Book¹⁶) and costs of medical care for health outcomes (obtained from the published literature). All costs were adjusted to year 2000 dollars using the medical component of the Consumer Price Index.¹⁷ Quality-adjusted life expectancy was calculated from utility values assigned to each health state in the model by 30 local internists. Because of the limitations of using physician utilities as proxies for patient utilities, sensitivity analyses were conducted using the range of relevant health state patient utilities reported in the literature.⁹ Future benefits, events, and costs were adjusted for time effects using a 3% discount rate.⁹

Because of limited randomized trial data and concerns about the generalizability of the data that are available, sensitivity analyses were performed to assess the impact of uncertainty of data inputs and to provide information for women with different risk profiles. One- and two-way sensitivity analyses were conducted to assess the impact of alternative assumptions about: 1) effectiveness of HRT in primary prevention of coronary heart disease; 2) effectiveness of raloxifene in primary prevention of coronary heart disease; 3) magnitude of breast cancer risk associated with HRT; 4) effectiveness of raloxifene in primary prevention of breast cancer; and 5) existence of any residual increase in risk of endometrial cancer with estrogen/progesterone regimens. For each sensitivity analysis, threshold values were identified where alternative regimens exceeded $50,000 per quality-adjusted life year and where alternatives no longer increased life expectancy. The range of values was taken from the widest 95% confidence interval in published studies or from the range of reasonable values developed through discussion with local experts. Because of uncertainty in the measurement of costs and utilities, the range for sensitivity analyses always included estimates from at least half to twice the base-case value.

Results

Compared with no treatment, both lifetime HRT and raloxifene therapy increase life expectancy and quality-adjusted life expectancy and are cost-effective for a 50-year-old postmenopausal woman at average risk for coronary heart disease and breast cancer. HRT provides an additional 0.65 discounted years of life expectancy at a net lifetime discounted cost of $3802 ($5849 per additional year of life); raloxifene an additional 0.71 discounted years of life expectancy at a net lifetime discounted cost of $12,968 ($18,265 per additional year of life) (Table 4). Because HRT reduces hip and vertebral fractures more than raloxifene therapy and fractures impact quality of life more than mortality, HRT increases quality-adjusted life years more than raloxifene therapy (gain of 1.75 versus 1.32 quality-adjusted life years) at a lower cost ($2173 versus $9824 per additional quality-adjusted life year). Thus, when choosing between lifelong raloxifene therapy and HRT, HRT is the dominant alternative (more effective and less costly). However, for shorter durations of therapy (ie, 5 or 10 years after menopause at age 50), raloxifene results in greater increase in life expectancy and quality-adjusted life expectancy than HRT at a cost of less than $50,000 per additional quality-adjusted life year.

As the estimated effectiveness of HRT for primary prevention of coronary heart disease declines, the relative effectiveness and cost-effectiveness of HRT decreases (Table 5). If the effect of HRT on lipid profiles from the Postmenopausal Estrogen/Progestin Interventions trial is used to estimate its impact on coronary heart disease, HRT decreases coronary heart disease risk by 25% (relative risk [RR] 0.75) and remains more effective and less expensive than raloxifene. If HRT does not reduce coronary heart disease risk, raloxifene becomes the preferred alternative with an incremental cost-effectiveness relative to HRT of $10,900 per quality-adjusted life year.

As the estimated effectiveness of raloxifene for primary prevention of coronary heart disease increases, raloxifene becomes relatively more effective and cost-effective than HRT. If raloxifene reduces coronary heart disease incidence by 30% (RR 0.70), raloxifene and HRT result in an equal gain in quality-adjusted life years. If the effect of raloxifene on coronary heart disease is equal to that estimated in the base-case for HRT (RR 0.5), raloxifene is the more cost-effective alternative.

As the risk of breast cancer from HRT increases, the relative effectiveness and cost-effectiveness of HRT compared with raloxifene decrease. However, HRT is both more effective and less expensive than raloxifene therapy across the range of published estimates (RR 0.9-1.74). If HRT does not increase the risk of breast cancer, use of HRT results in an increase of 0.85 quality-adjusted life years compared with use of raloxifene at a cost saving of $10,900.

As raloxifene becomes more effective in primary prevention of breast cancer, it becomes relatively more effective and cost-effective than HRT. If raloxifene reduces the incidence of breast cancer by 90% (RR 0.1), raloxifene results in a gain in 1.66 quality-adjusted life years compared with no therapy. However, if one assumes coronary heart disease risk reduction from HRT, this gain in quality-adjusted life expectancy is still less than that seen with HRT. If raloxifene is less effective in primary prevention of breast cancer than estimated in the base-case analysis (RR 0.36 or higher), the relative benefit of HRT further increases.

The risk of endometrial cancer from HRT has little substantive effect on the relative benefit of HRT. If HRT increases the risk of endometrial cancer four-fold (RR 4.0), the incremental gain in quality-adjusted life years for HRT compared with raloxifene therapy falls to 0.07, but HRT remains both more effective and less expensive.

The relative benefit of these therapies depends upon a woman's risk of coronary heart disease, osteoporosis, and breast cancer. Because the benefit of HRT in reducing coronary heart disease and osteoporosis risk is believed to be substantially greater than that of raloxifene, HRT remains the more effective and less expensive alternative for women at increased risk of coronary heart disease and osteoporosis. However, increases in breast cancer risk have a significant impact on the relative benefit of raloxifene and HRT (Table 6). If a woman has a 40% increase over the estimated population lifetime breast cancer risk of 10% (ie, lifetime risk of 14%), raloxifene results in an equal gain in quality-adjusted life expectancy as HRT, and HRT is no longer the dominant alternative. If a woman has a 50% increase in breast cancer risk (ie, lifetime breast cancer risk of 15% or higher), raloxifene becomes the more cost-effective alternative.

Variation in the estimates of costs, utilities, and discount rates has little substantive effect on which alternative therapy is preferred. If the cost of raloxifene falls to $175 per year, raloxifene becomes the less costly alternative ($12,496 versus $12,518). However, HRT still results in a greater gain in quality-adjusted life expectancy with an incremental cost-effectiveness ratio of $51 per quality-adjusted life year compared with raloxifene. HRT remains the dominant or cost-effective alternative for a woman at average coronary heart disease and breast cancer risk across the ranges of costs examined for HRT, coronary disease, breast cancer, osteoporosis, endometrial cancer, or thromboembolism. Furthermore, although the relative benefit of HRT decreases as the discount rate decreases, if neither costs nor life years are discounted, HRT remains the preferred option, with an incremental cost-effectiveness ratio of raloxifene compared with HRT of $882,896 per quality-adjusted life year.

Although the relative benefit of HRT decreases as the utility estimates for coronary heart disease and osteoporosis increase and the estimates for breast cancer decrease, HRT remains the dominant or cost-effective alternative across the range of utility estimates examined. Because HRT reduces menopausal symptoms whereas raloxifene does not, and this issue may be particularly relevant for women taking therapy for only 5 or 10 years after menopause, we examined the effect of an improvement in utility with HRT compared with raloxifene for these time frames. For short-term therapy, if the model assumes even modest benefit in quality of life from HRT compared with raloxifene (absolute increase of 2% or higher), HRT is both more effective and less expensive than raloxifene therapy for 5- to 10-year courses of therapy.

Discussion

Because of the availability of alternative hormonally active therapies that differ in their impact on coronary heart disease, breast cancer, and osteoporotic fracture, and increasing controversy about the effects of HRT on coronary heart disease, we performed a decision analysis to estimate the clinical (life expectancy and quality-adjusted life expectancy) and economic (incremental cost-effectiveness) impact of HRT and raloxifene in postmenopausal women. Assuming the benefit of HRT on coronary risk reported in observational studies and the benefit of raloxifene on coronary risk extrapolated from its effects on lipids, both long-term HRT and long-term raloxifene increase both life expectancy and quality-adjusted life expectancy in 50-year-old postmenopausal women at average risk for coronary heart disease and breast cancer. Because HRT increases quality-adjusted life expectancy more than raloxifene, and raloxifene is more costly than HRT, HRT is the dominant (more effective and less costly) alternative in this setting. Thus, despite raloxifene's apparent reduction in breast cancer incidence, long-term HRT remains the most cost-effective therapy for women at average breast cancer risk seeking treatment to increase their quality-adjusted life expectancy after menopause.

The relative benefits of raloxifene and HRT depend upon a woman's breast cancer risk. For a woman with a predicted lifetime 50% increase in breast cancer risk (ie, lifetime risk of 15% or higher), raloxifene is a cost-effective alternative to HRT, resulting in a greater increase in quality-adjusted life expectancy at an incremental cost of less than $50,000 per quality-adjusted life year. The most widely used and validated model for individual breast cancer risk prediction is the Gail model.^18,19 Gail model software can be obtained from the National Cancer Institute at 1-800-4CANCER or http://cancertrials.nci.nih.gov/forms/CtRiskDisk.html. If using a software program is not feasible, certain breast cancer risk factors (one or more first-degree relatives with breast cancer, one or more prior breast biopsies, history of atypical hyperplasia on a breast biopsy, and carrying a mutation in BRCA1 or BRCA2) consistently convey an RR of breast cancer over 1.5 and can be used to identify women who have a 15% or greater lifetime risk of breast cancer.¹⁹

The relative benefits of raloxifene and HRT also change significantly with alternative assumptions about the effects of HRT on coronary heart disease risk. If HRT proves to reduce the risk of a first coronary heart disease event by less than 20%, long-term raloxifene becomes the more cost-effective alternative for all women. If the effects of both HRT and raloxifene are extrapolated from changes in lipids, HRT remains the more cost-effective alternative.^5,13 These results provide evidence to help clinicians interpret and implement recent American Heart Association guidelines that suggest decisions about HRT in women without cardiovascular disease "should be based on established noncoronary benefits and risks, possible coronary benefits and risks, and patient preference."²⁰

For women interested in pharmacologic therapy for 5 or 10 years after menopause, raloxifene is associated with a greater increase in life expectancy and quality-adjusted life expectancy than HRT at a cost of less than $50,000 per quality-adjusted life year. A woman's risk of death from breast cancer compared with her risk of death from coronary disease and osteoporosis is relatively greater at younger than older ages. Thus, raloxifene's reduction of breast cancer risk has its greatest impact in the years immediately after menopause. However, the beneficial effect of HRT on menopausal symptoms was not included in this analysis. Even a relatively small symptomatic benefit of HRT relative to raloxifene results in a greater increase in quality-adjusted life years with short-term HRT than with short-term raloxifene.

These results extend prior research in this area. Previous decision analyses without discounting have found HRT to increase life expectancy by 0.5 to 1 year in average-risk women.^21-24 In this analysis, HRT increased life expectancy by 1.0 years in the absence of discounting. One cost-effectiveness analysis also found HRT to be cost-effective compared with no therapy.²¹ A recently published decision analysis of alendronate, raloxifene, and HRT found that raloxifene increased life expectancy more than HRT for women at high breast cancer risk and low coronary heart disease risk.²⁴ However, this prior analysis did not include the recent data about the benefit of raloxifene on breast cancer risk in the base-case analysis or the effects of raloxifene and HRT on vertebral fractures or thrombosis. Furthermore, the current study is the first to assess the comparative economic impact of alternative therapies.

The current study has several limitations. We chose to focus on hormonally active options for postmenopausal women because these options have many competing effects, making a decision analysis particularly valuable. We did not include the many other options for prevention of osteoporosis, coronary disease, and breast cancer that have a single main effect (eg, statins, alendronate), and that may be even more effective than either HRT or raloxifene for a specific complication of hormonal deficiency. However, deciding between options for prevention of a single disorder is potentially less complex, and including all options would make the current analysis difficult to use. Because both HRT and raloxifene have side effects, and an extensive literature search found no evidence that patient adherence differs between the therapies, we did not include the effects of noncompliance in the model. In addition, for many of the model parameters, only limited data are currently available. For example, data on the impact of raloxifene on breast and endometrial cancer come from a single large clinical trial.⁴ Although we used the best available evidence for each model parameter estimate, uncertainty is inevitable (eg, effect of HRT on coronary heart disease). In this setting, sensitivity analyses were used to understand the impact of the ranges of uncertainty and provide an important context for understanding the base-case results.

Postmenopausal women now have several options to reduce their long-term risk of coronary heart disease, osteoporosis, and breast cancer. This analysis suggests that for the great majority of postmenopausal women without a major breast cancer risk factor, long-term HRT remains the dominant alternative, resulting in a greater increase in quality-adjusted life expectancy at a lower cost. However, long-term raloxifene therapy is a cost-effective alternative for postmenopausal women at significantly increased risk of breast cancer and is a cost-effective alternative for women with average breast cancer risk who will not take HRT. Until the results of large scale randomized trials of HRT as primary prevention become available, women and physicians continue to face difficult decisions about postmenopausal therapy. This analysis provides important evidence to make more informed decisions and may make counseling postmenopausal women a little easier.

References

1. Grodstein F, Stampfer M, Colditz GA, Willett WC, Manson JE, Joffe M, et al. Postmenopausal hormone therapy and mortality. N Engl J Med 1997;336:1769-75.

2. Folsom AR, Mink PJ, Sellers TA, Hong CP, Zheng W, Potter JD. Hormonal replacement therapy and morbidity and mortality in a prospective study of postmenopausal women. Am J Public Health 1995;85:1128-32.

3. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: Collaborative reanalysis of data from 51 epidemiologic studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet 1997;350:1047-59.

4. Cummings SR, Eckert S, Krueger KA, Grady D, Powles TJ, Cauley JA, et al. The effects of raloxifene on risk of breast cancer in postmenopausal women: Results from the MORE randomized trial. JAMA 1999;281:2189-97.

5. Delmas PD, Bjarnson NH, Mitlak BH, Ravoux AC, Shah AS, Huster WJ, et al. Effects of raloxifene on bone mineral density, serum cholesterol concentrations, and uterine endometrium in postmenopausal women. N Engl J Med 1997;337:1641-7.

6. Walsh BW, Kuller LH, Wild RA, Paul S, Farmer M, Lawrence JB, et al. Effects of raloxifene on serum lipids and coagulation factors in healthy postmenopausal women. JAMA 1998;279:1445-51.

7. Ettinger B, Black DM, Mitlak BH, Knickerbocker RK, Nickelsen T, Genant HK, et al. Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene. JAMA 1999;282:637-45.

8. Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, et al. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. JAMA 1998;280:605-13.

9. Gold MR, Siegel JE, Russell LB, Weinstein MC, eds. Cost-effectiveness in health and medicine. New York, NY: Oxford University Press, 1996.

10. Murphy SL. Deaths: Final data for 1998. National vital statistics reports, vol. 48, no. 11. Hyattsville, MD: National Center for Health Statistics, 2000.

11. US Department of Health, Education, and Welfare. Some risk factors related to the annual incidence of cardiovascular disease and death using pooled repeated biennial measurements: Framingham heart study, 30-year follow-up. In: The Framingham study: An epidemiological investigation of cardiovascular disease, Section 34. Bethesda, MD: National Heart, Lung, and Blood Institute, 1987.

12. Kinosian B, Glick H, Garland G. Cholesterol and coronary heart disease: Predicting risks by levels and ratios. Ann Intern Med 1994;121:641-7.

13. Writing Group for the PEPI Trial. Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women: The Postmenopausal Estrogen/Progestin Interventions (PEPI) trial. JAMA 1995;273:199-208.

14. Brett KM, Madans JH. Long term survival after coronary heart disease: Comparisons between men and women in a national sample. Ann Epidemiol 1995;5:25-32.

15. Rosamond WD, Chambless LE, Folsom AR, Cooper LS, Conwill DE, Clegg L, et al. Trends in the incidence of myocardial infarction and in mortality due to coronary heart disease, 1987 to 1994. N Engl J Med 1998;339:861-7.

16. 1999 Red Book. Montvale, NJ: Medical Economics Data, 1999.

17. Bureau of Labor Statistics. Available via the Internet at http://stats.bls.gov/. Last accessed October 17, 2001.

18. Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 1989;81:1879-86.

19. Armstrong K, Eisen A, Weber B. Assessing the risk of breast cancer. N Engl J Med 2000;324:564-71.

20. American Heart Association. Estrogen and cardiovascular disease in women. Available via the Internet at http://www.americanheart.org/Heart_and_Stroke_A_Z_Guide/estrogen.html.

21. Weinstein MC. Estrogen use in postmenopausal women — costs, risks, and benefits. N Engl J Med 1980;303:308-16.

22. Grady D, Rubin SM, Petitti DB, Fox CS, Black D, Ettinger B, et al. Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med 1992;117:1016-37.

23. Gorsky RD, Koplan JP, Peterson HB, Thacker SB. Relative risks and benefits of long-term estrogen replacement therapy: A decision analysis. Obstet Gynecol 1994;83:161-6.

24. Col NF, Pauker SG, Goldberg RJ, Eckman MH, Orr RK, Ross EM, et al. Individualizing therapy to prevent long-term consequences of estrogen deficiency in postmenopausal women. Arch Intern Med 1999;159:1458-66.

25. Farmer ME, White LR, Brody JA, Bailey KR. Race and sex differences in hip fracture incidence. Am J Public Health 1984;74:1374-80.

26. Lu-Yao GL, Baron JA, Barrett JA, Fisher ES. Treatment and survival among elderly Americans with hip fractures: A population based study. Am J Public Health 1994;84:1287-91.

27. Cooper C, Atkinson EJ, O'Fallon WM, Melton LJ III. Incidence of clinically diagnosed vertebral fractures: A population-based study in Rochester, Minnesota, 1985-1989. J Bone Miner Res 1992;7:221-7.

28. Ries LAG, Kosary CL, Hankey BF, Harras A, Miller BA, Edwards BK, eds. SEER cancer statistics review, 1973-1993: Tables and graphs. Bethesda, MD: National Cancer Institute, 1996.

29. Grodstein F, Stampfer MJ, Goldhaber SZ, Manson JE, Colditz GA, Speizer FE, et al. Prospective study of exogenous hormones and risk of pulmonary embolism in women. Lancet 1996;348:983-7.

30. Voigt LF, Weiss NS, Chu JR, Daling J, McKnight B, van Belle G. Progestagen supplementation of exogenous oestrogens and risk of endometrial cancer. Lancet 1991;338:274-7.

31. Kannel WB, Vokonas PS. Demographics of the prevalence, incidence, and management of coronary heart disease in the elderly and in women. AEP 1992;2:5-14.

32. Russell MW, Huse DM, Drowns SH, Hamel EC, Hartz SC. Direct medical costs of coronary artery disease in the United States. Am J Cardiol 1998;81:1110-5.

33. Taplin SH, Barlow W, Urban N, Mandelson MT, Timlin DJ, Ichikawa L, et al. Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care. J Natl Cancer Inst 1995;87:417-26.

34. Brainsky A, Glick H, Lydick E, Epstein R, Fox K, Hawkes W, et al. The economic cost of hip fracture in community-dwelling older adults: A prospective study. J Am Geriatr Soc 1997;45:281-7.

35. DeLaet CE, van Hout BA, Burger H, Weel AE, Hofman A, Pols HA. Incremental cost of medical care after hip fracture and first vertebral fracture: The Rotterdam study. Osteopor Int 1999;10:66-72.

36. Feldman S, Berkowitz RS, Tosteson AN. Cost-effectiveness of strategies to evaluate postmenopausal bleeding. Obstet Gynecol 1993;81:968-75.

37. Hull RD, Raskob GE, Rosenbloom D, Pineo GF, Lerner RG, Gafni A, et al. Treatment of proximal vein thrombosis with subcutaneous low-molecular-weight heparin vs intravenous heparin: An economic perspective. Arch Intern Med 1997;157:289-94.

Address reprint requests to: Katrina Armstrong, MD, MSc, University of Pennsylvania, 1233 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021; E-mail: karmstro@mail.med.upenn.edu

Received February 22, 2001.
Received in revised form August 10, 2001.
Accepted August 16, 2001.