Osteoporotic fractures are a significant public health problem, resulting in substantial morbidity and mortality. Previous estimates of the economic burden of osteoporosis, however, have not fully accounted for the costs associated with treatment of nonhip fractures, minority populations, or men. Accordingly, the 1995 total direct medical expenditures for the treatment of osteoporotic fractures were estimated for all persons aged 45 years or older in the United States by age group, sex, race, type of fracture, and site of service (inpatient hospital, nursing home, and outpatient). Osteoporosis attribution probabilities were used to estimate the proportion of health service utilization and expenditures for fractures that resulted from osteoporosis. Health care expenditures attributable to osteoporotic fractures in 1995 were estimated at $13.8 billion, of which $10.3 billion (75.1%) was for the treatment of white women, $2.5 billion (18.4%) for white men, $0.7 billion (5.3%) for nonwhite women, and $0.2 billion (1.3%) for nonwhite men. Although the majority of U.S. health care expenditures for the treatment of osteoporotic fractures were for white women, one-fourth of the total was borne by other population subgroups. By site-of-service, $8.6 billion (62.4%) was spent for inpatient care, $3.9 billion (28.2%) for nursing home care, and $1.3 billion (9.4%) for outpatient services. Importantly, fractures at skeletal sites other than the hip accounted for 36.9% of the total attributed health care expenditures nationally. The contribution of nonhip fractures to the substantial morbidity and expenditures associated with osteoporosis has been underestimated by previous researchers.
Osteoporosis is recognized as an important public health problem because of the significant morbidity and mortality associated with its complications, namely fractures of the proximal femur (hip), vertebrae (spine), distal forearm, proximal humerus, pelvis, and other skeletal sites.1 Epidemiologically, elderly white women have the highest incidence of osteoporotic fractures, whereas black women and all men are at lower risk.2–5 Seely et al. demonstrated that fractures of most sites in the elderly are related to osteoporosis.6,7 Compared with other osteoporotic fractures, however, fractures of the hip incur the greatest morbidity and direct medical costs for health services.1 Hip fractures cause hospitalization, disability, and loss of independence for an estimated 300,000 persons annually.8 Overall, the case fatality rate for hip fracture patients within 1 year following the fracture is 24%,9–12 and survivors frequently experience sustained disability, often leading to institutionalization.13–18 Fractures at skeletal sites other than the hip have received much less attention even though they can also cause substantial disability.19–22
Previous studies documenting the economic consequences of treating complications of osteoporosis have primarily been restricted to either hip fractures or female study populations, resulting in an underestimate of the burden of this disease. The direct medical costs associated with hip fractures among men and women in the United States have been estimated to be between 5.4 and $7.1 billion.8,23,24 Phillips et al. estimated the direct medical costs of osteoporotic fractures for women aged 45 years and older at $5.2 billion nationally in 1986.25 Chrischilles et al. estimated the 10-year direct medical costs for hip, spine, and forearm fractures among white women aged 45 years and older at $45.2 billion.26 Holbrook et al., in the most comprehensive study to date, included men and women as well as fractures of the hip and other skeletal sites to estimate the direct medical costs of osteoporosis at $5.2 billion in the United States in 1984.23 There has been no recent assessment since Holbrook et al.23 of the contribution of men, minority populations, and nonhip fractures to the morbidity and economic consequences of osteoporosis.
The goal of this prevalence-based cost-of-illness study is to estimate the direct medical expenditures associated with treating osteoporotic fractures for all persons aged 45 years or older in the United States in 1995. Our estimates encompass fractures of the hip, spine, and forearm as well as other selected sites such as the humerus, ribs, pelvis, etc. This study is unique in assessing the contribution of nonwhites, males, and all nonhip fractures to the total medical expenditures associated with osteoporotic fractures. The results are intended to provide a frame of reference for evaluating the cost-effectiveness of public health measures to prevent osteoporotic fractures as well as programs that screen high-risk individuals for osteoporosis and new technologies used in the diagnosis and treatment of osteoporosis.
MATERIALS AND METHODS
Attributing fractures to osteoporosis
Since all fractures relate both to the skeletal load applied and to bone strength,27,28 it is not possible using health survey data to ascertain whether a given fracture is due to trauma or to the skeletal fragility that results from osteoporosis. Furthermore, such data are not available in the medical literature. Since osteoporosis, like other chronic conditions, is dramatically under-reported as a comorbidity on hospital discharge abstracts,29–33 health care expenditures associated with osteoporotic fractures would be grossly underestimated by attributing only those expenditures for health care encounters where osteoporosis was reported as a secondary diagnosis. Consequently, to estimate the proportion of fractures at specific sites that are related to osteoporosis, criteria previously developed through a modified Delphi consensus technique by an Expert Panel were used.34 The Expert Panel consisted of six clinicians from the fields of internal medicine, endocrinology, rheumatology, orthopedic surgery, and nuclear medicine with extensive expertise in the diagnosis and treatment of patients with osteoporosis. In a three-stage Delphi process, the panelists assessed the contribution of osteoporosis to the development of fractures of the hip, spine, forearm, and all other sites by determining osteoporosis attribution probabilities for various patient populations according to sex, age (45–64, 65–84, ≥85 years), and race (white, black, all others). The specific fractures included in each of the four classes and their corresponding medical diagnosis codes are available from the corresponding author (N.F.R.). To document the degree of uncertainty associated with each attribution probability, the panelists ranked the final attribution probabilities using the following validity score: 1 = certain, low risk of the attribution probability being wrong (±5% of the osteoporosis attribution probability); 2 = reliable, some risk of being wrong (±10%); 3 = risky, substantial risk of being wrong (±20%); or 4 = unreliable, great risk of being wrong (more than 20%).
For the purposes of this analysis, it was necessary to combine selected attribution probabilities developed by the Expert Panel into broader groupings (e.g., nonhip fractures of white and non-white persons and hip fractures of non-white persons). The attribution probabilities for these broader groups were derived by calculating the median value across the selected demographic strata and fracture sites. For example, to determine the attribution probability for nonhip fractures among non-white populations used in this study, the median of the reported attribution probabilities for fractures of the spine, forearm, and all other sites for persons who are black or other minorities was computed. The corresponding validity scores associated with the attribution probabilities for each of these groups was determined by computing the mean value across the corresponding fracture site, age, sex, and race strata. The osteoporosis attribution probabilities and corresponding validity scores used in this analysis are presented in Appendix 1.
Estimation of health resource utilization
In this study, national health care survey data from the National Center for Health Statistics and the Agency for Health Care Policy and Research were employed to estimate health resource utilization and expenditures attributable to osteoporotic fractures for persons aged 45 years and older in the U.S. in 1995. An overview of each data source used in this study is presented in Table 1. Health care encounters of persons with a primary and/or secondary diagnosis of osteoporotic fractures were identified by the corresponding diagnosis codes for each fracture site. The remainder of this section describes the methods used to estimate health resource utilization and expenditures by site-of-service.
Table Table 1. Sources of Data for Osteoporotic Fracture Utilization and Expenditures Estimates
Inpatient hospital services: Data used to estimate inpatient hospital utilization were extracted from the 1992 National Hospital Discharge Survey. All hospital discharges with a primary or secondary medical diagnosis of an osteoporotic fracture were identified. Race was not stated for 20.1% of the discharges identified. To account for records with unknown race, estimates of the number of hospitalizations by race were adjusted according to the distribution of known race values.35 For hospitalizations with a primary diagnosis of an osteoporotic fracture, inpatient days (i.e., number of days of care) were computed for each fracture site–age-sex-race group. In estimating the total inpatient days attributable to osteoporotic fractures, the corresponding osteoporosis attribution probabilities for each fracture site–age-sex-race category (Appendix 1) were applied to the number of hospitalized days. By contrast, only the excess days for hospitalization with a secondary osteoporotic fracture diagnosis were attributed to osteoporotic fractures. For each fracture site–age-sex-race category, the difference in the length of hospital stay for patients with a secondary diagnosis of an osteoporotic fracture and patients without any mention of an osteoporotic fracture was determined. The corresponding osteoporosis attribution probabilities were then applied to the difference in days of care in each fracture site–age-sex-race stratum.
Nursing home care: Two data sources were used to estimate the total number of nursing home residents with osteoporotic fractures. First, the Discharge Resident file of the 1985 National Nursing Home Survey was used to determine the proportion of osteoporotic fracture patients admitted from short-term hospitals, other medical facilities (e.g., domiciliary or personal care facilities, intermediate care facilities, skilled nursing facilities, veteran's hospitals, long-term hospitals), or private residences. Of the nursing home residents with a primary admission diagnosis of an osteoporotic fracture, 60.6 and 74.7% of men and women, respectively, were admitted from short-term hospitals compared with 39.4 and 25.3% of men and women, respectively, who were admitted from other medical facilities or private residences. A more current data source, the 1992 National Hospital Discharge Survey, was used to estimate the actual number of osteoporotic fracture patients who were admitted to nursing homes from short-term hospitals. From this latter survey, all inpatients with a primary diagnosis of an osteoporotic fracture who were discharged to long-term care institutions were selected. The total number of nursing home admissions was estimated by adding the number of residents admitted from short-term hospitals and the number of residents admitted from other medical facilities and private residences, which was estimated using the known proportions of each population. The corresponding osteoporosis attribution probabilities for each fracture site–age-sex-race stratum were then applied to estimate the number of nursing home stays attributable to osteoporotic fractures.
The total number of long-term care days associated with osteoporotic fractures was derived from the 1985 National Nursing Home Survey. The mean length of nursing home stays was computed for discharged residents with a primary admission diagnosis of an osteoporotic fracture. The mean length of stay was stratified for residents originally admitted from short-term hospitals and those admitted from other medical facilities and private residences because of the known association between nursing home duration and admission source.36 Among nursing home residents with osteoporotic fractures, the mean length of stay for those admitted from short-term hospitals was 120 days for men and 205 days for women, and the mean length of stay for those admitted from other medical facilities or private residences was 331 and 469 days, respectively. Because only expenditures for 1995 are included in this study, nursing home stays with a mean length of stay of more than 365 days were censored at 365 days. The number of long-term care days attributable to osteoporotic fractures was determined by multiplying the mean length of a nursing home stay by the total number of attributed nursing home residents for each fracture site–age-sex-race stratum.
Outpatient services: The outpatient health care utilization for patients with osteoporotic fractures was estimated for the following services: office-based physician care, outpatient hospital and emergency room care, ambulance services, diagnostic radiology services, home health care, home medical supplies, over-the-counter and prescription medications, and therapeutic treatments. The number of office-based outpatient physician visits, as well as the associated diagnostic radiology services, physical therapy services, and prescription and nonprescription medications ordered during these encounters, were estimated using the 1992 National Ambulatory Medical Care Survey. Ambulatory medical care received by patients with osteoporotic fractures through hospital outpatient and emergency room departments, including diagnostic radiology services, physical therapy, and prescription and nonprescription medications, was estimated using the Outpatient Department and Emergency Department files of the 1992 National Hospital Ambulatory Medical Care Survey. The Medical Equipment Purchases and Rentals file of the 1987 National Medical Expenditure Survey was employed to estimate the use of orthopedic equipment (crutches, wheelchairs, walkers, corrective shoes, etc.), other medical items (special home or car alterations or equipment, bathing equipment, beds, etc.), and ambulance services by patients with osteoporotic fractures. The Current and Discharged Patient files of the 1992 National Home and Hospice Care Survey were used to estimate home health service utilization among patients with osteoporotic fractures.
From each of the data sources listed above, all outpatient encounters with a primary diagnosis of an osteoporotic fracture were identified, and the corresponding osteoporosis attribution probabilities were applied to estimate the outpatient care attributable to osteoporotic fractures. To correct for the missing race values from 22.4% of current and 15.7% of discharged home health care patients with osteoporotic fractures, the same methodology used to redistribute the “not stated” race in the inpatient hospital analysis was employed.
Estimation of 1995 utilization levels: The data sources used in this study reflect health resource utilization in 1985 through 1992. To estimate 1995 health resource utilization, each utilization measure for each age-sex-race group was inflated by the combined increases in the civilian population of the United States from the year represented by the utilization data to 1995.37
Estimation of medical expenditures
Total medical expenditures for osteoporotic fractures consist of costs associated with short-term hospital care, nursing home care, outpatient office–based physicians' services, hospital outpatient services, emergency room services, ambulance services, and use of drugs and medical equipment. Data from the 1987 National Medical Expenditure Survey were used to estimate mean expenditures for encounters with a primary diagnosis of an osteoporotic fracture for each of the sites of service listed above. Table 1 presents the mean expenditures per unit of service used in this study. Expenditures were based on payments from Medicare, Medicaid, private third-party payers, and other sources. The cost per inpatient hospital day represents the sum of facility and physician expenses. The Consumer Price Indices for Hospital and Related Services, Physician Services, Medical Care Services, Professional Medical Services, Drugs, and Medical Care Commodities were used to inflate unit expenditures from 1987 to 1995 dollars.38 The Health Care Financing Administration's Nursing Facility Input Price Index was used to inflate nursing home expenditures to 1995 levels.39 Expenditures associated with osteoporotic fractures were estimated by multiplying the utilization attributable to osteoporotic fractures by the mean expenditure per unit of service in 1995 dollars. Total health expenditures for the treatment of osteoporotic fractures in 1995 were obtained by summing across the three types of services included in this analysis.
Health resource utilization attributable to osteoporotic fractures
Utilization of hospital, nursing home, and outpatient health care services attributable to osteoporotic fractures in 1995 are shown in Tables 2 and 3. Of the 432,448 hospitalizations with a primary diagnosis of an osteoporotic fracture for persons aged 45 years and older, 57.0% were for hip fractures, 6.8% were for spine fractures, 3.1% were for forearm fractures, and the remaining 33.0% were for treatment of fractures at other sites. In terms of patient demographics, 78.9% of all hospitalizations for osteoporotic fractures were for women, and of these, 6.4% were for non-white women. An estimated 4.1 million days were associated with these stays, averaging 9.6 days per hospital stay. Hospital days associated with stays with a primary diagnosis of a hip fracture constituted 70.6% of total attributed hospital days. On average, inpatients with a secondary diagnosis of an osteoporotic fracture remained hospitalized 4.4 days longer than those without mention of an osteoporotic fracture (data not shown). Hospitalizations with a secondary osteoporotic fracture diagnosis contributed an additional 509,136 days to total hospital utilization for osteoporotic fractures. Of these stays, 62.1% were for treatment of fractures at skeletal sites other than the proximal femur, forearm, and spine.
Table Table 2. Health Resource Utilization Attributable to Osteoporotic Fractures in the United States, by Type of Service, Sex, Race, and Type of Fracture, 1995
Table Table 3. Distribution of Health Resource Utilization Attributable to Osteoporotic Fractures in the United States, by Type of Service and Type of Fracture, 1995
Altogether, 179,221 nursing home stays associated with 44.6 million patient days were attributed to osteoporotic fractures in 1995. Hip fractures accounted for 76.9% of these stays and 72.5% of the attributed days. Three-quarters (75.9%) of the nursing home residents were white women, of whom 70.6% resided in a nursing home due to a hip fracture. Of the remaining residents, 4.4% were non-white women and 19.7% were men, contributing an additional 9.1 million days to total nursing home utilization for osteoporotic fractures.
A total of 3.4 million outpatient physician, outpatient hospital, and emergency room examinations were attributed to osteoporotic fractures in 1995; nearly two-thirds (63.7%) of these outpatient examinations were for the treatment of fractures at skeletal sites other than the hip, forearm, and spine. These outpatient encounters were disproportionately for white women (75.6%), followed by men (19.5%), and non-white women (4.9%). Diagnostic imaging and physical therapy services were provided during 55.2% and 5.6%, respectively, of the outpatient examinations associated with osteoporotic fractures. In addition, a total of 2.4 million prescription and nonprescription medications were prescribed during these outpatient examinations. Nearly 220,000 ambulance encounters were attributed to osteoporotic fractures in 1995, and nearly half a million orthopedic and other medical supplies were provided to treat osteoporotic fractures. Finally, approximately 2.2 million home health care visits were attributed to osteoporotic fractures.
While outpatient care was provided predominately for the treatment of fractures at skeletal sites other than the hip, forearm, and spine, there are several notable exceptions. First, 59.5% of home health care encounters were for the treatment of hip fractures. Second, more than half of the 193,557 physical therapy sessions attributable to osteoporosis were either for the treatment of hip fractures (22.9%) or forearm fractures (38.7%). Finally, nearly half of all ambulance encounters attributable to osteoporosis (45.3%) were due to either fractures of the hip or spine.
Health care expenditures attributable to osteoporotic fractures
Health care expenditures attributable to osteoporotic fractures in 1995 were estimated at $13.76 billion, of which $10.34 billion (75.1%) was for the treatment of white women, $2.53 billion (18.4%) for the treatment of white men, $0.73 billion (5.3%) for the treatment of non-white women, and $0.17 billion (1.3%) for the treatment of non-white men (Table 4). Health care expenditures were greatest for patients aged 65–84 years (52.8%), followed by patients 85 years and older (34.8%), and patients aged 45–64 years of age (12.4%). Expenditure estimates by type of service include $8.60 billion (62.4%) for hospitalization, $3.88 billion (28.2%) for nursing home care, and $1.30 billion (9.4%) for outpatient services. Expenditures for outpatient services were primarily associated with care provided in emergency rooms (43.8%) and physicians' offices (36.3%) (Table 5). By type of service and age, inpatient expenditures attributed to osteoporotic fractures were greatest for patients aged 65–84 years; nursing home expenditures were greatest for patients 85 years and older; and outpatient expenditures were greatest for patients aged 45–64 years (data not shown).
Table Table 4. Health Care Expenditures Attributable to Osteoporotic Fractures in the United States, by Type of Service, Age, Race, and Type of Fracture, 1995
Table Table 5. Health Care Expenditures Attributable to Osteoporotic Fractures in the United States by Type of Service and Type of Fracture, 1995
Treatment of osteoporotic hip fractures accounted for $8.68 billion (63.1%), and fractures at other sites accounted for $5.08 billion (36.9%) of total health care expenditures attributed to osteoporotic fractures (Table 5). Not surprisingly, the distribution of expenditures by type of service differed between fractures of the hip and fractures at other skeletal sites. Whereas 72.9% of inpatient and nursing home expenditures attributable to osteoporosis were for the treatment of fractures of the hip and spine, only 26.3% of attributable outpatient expenditures were for the treatment of these fracture sites. Additionally, the distribution of expenditures by site of fracture differed between service types. For example, the proportion of expenditures devoted to inpatient and nursing home care varied from 58.2% for forearm fractures to 79.9% for fractures at skeletal sites other than the hip, forearm, and spine and to approximately 95% of expenditures for fractures of the spine and hip.
Sensitivity analysis of osteoporosis attribution probability estimates
A sensitivity analysis was conducted to ascertain the effect of varying the osteoporosis attribution probabilities using the uncertainty measures provided by the Expert Panelists. The variability associated with the health resource utilization and expenditure measures used in this study were not included in the sensitivity analysis since previous studies have reported on the appropriateness and reliability of survey data from the National Center for Health Statistics and the Agency for Health Care Policy and Research.40–42 To conduct the sensitivity analysis, the lower and upper bounds for each osteoporosis attribution probability were calculated by multiplying the uncertainty percentages associated with each validity ranking reported by the Expert Panel. Thus, taking into account the degree of confidence with which the panel estimated each attribution probability, health care expenditures for osteoporotic fractures at all sites are estimated to vary by 9.1%, ranging from 12.51 to $15.02 billion (Table 6). For hip fractures, health care expenditures are estimated to vary by 6.6%, ranging from 8.11 to $9.26 billion. The relatively narrow ranges indicate that the uncertainty of the Expert Panel regarding the osteoporosis attribution probabilities did not have a major impact on the resultant expenditure estimates.
Table Table 6. Sensitivity Analysis of Osteoporosis Attribution Probability Estimates
Using prevalence-based cost-of-illness methods and data from national health care surveys, health care expenditures associated with all osteoporotic fractures for persons aged 45 years and older in the United States in 1995 were estimated. A study of the cost of a given illness can provide insights into the pattern of expenditures for that disease, the patient case-mix, and the categories of spending that might be modified to reduce expenditures. In the case of osteoporosis, inpatient services comprise the largest single component of medical expenditures, followed by nursing home care. Together they comprise more than 90% of all expenditures for osteoporotic fractures in 1995. In terms of patient case-mix, treatment for osteoporotic fractures is predominately for the treatment of whites, who accounted for 93% of all attributable medical expenditures.
Several prevalence-based cost-of-illness studies estimating medical expenditures for hip and other osteoporotic fractures have been conducted during the past two decades.8,23–26 To permit comparison of the various study results to our estimates, total direct medical expenditures derived in each of these studies were inflated to 1995 dollars using the Bureau of Labor Statistics' Consumer Price Index for Medical Care.38 As shown in Table 7, total 1995 direct medical expenditures for osteoporotic fractures range from $9.4 billion (Phillips et al.25) to $10.7 billion (Holbrook et al.23) to $13.8 billion (this study). Total 1995 direct medical expenditures for hip fractures range from $7.3 billion (OTA Study8) to $8.7 billion (this study) to $9.3 billion (Praemer et al.24) and $12.4 billion (Holbrook et al.23) It is important to note that the other studies estimated expenditures for all hip fractures, whereas this study included only hip fracture costs attributable to osteoporosis.
Table Table 7. Comparison of U.S. Osteoporotic Fracture Cost-of-Illness Studies*
National prevalence-based cost-of-illness studies, including this analysis, generate results that do not provide a framework for assessing costs on a per patient basis, although the aggregate costs are often impressive in magnitude. Consequently, per patient 1995 hip fracture expenditures were estimated by dividing the total direct medical costs of hip fractures, derived in each of the above studies, by the total incident hip fracture population in the U.S. in 1995. The total incident hip fracture population, estimated at 267,733 persons aged 45 years and older, was calculated by multiplying age- and sex-specific hip fracture incidence rates for Olmsted County, Minnesota, residents in 1985–199243 by the corresponding age and sex distribution of the U.S. civilian population in 1995. Although limited to a small geographic area, these data were used because national hip fracture incidence rates for persons aged 45 years and older have not been reported since 1970–1983.44 As shown in Table 5, 1995 per patient expenditures for hip fractures ranged from $27,318 (OTA Study)8 to $32,428 (this study) to $34,696 (Praemer et al.)24 and $35,104 (Holbrook et al.).23 For comparison, two alternate data sources, Ho et al.45 and the 1992 National Hospital Discharge Survey,46 were used to estimate the total number of hip fractures nationally. The resultant per patient hip fracture expenditures based on these varying population estimates are $30,850 and $33,428, respectively, using cost data from this study.
The variation in per patient costs, most notably the lower costs of the OTA study, is due in part to differing methodologies used in each study. In the present study, as well as those by Praemer et al.24 and Holbrook et al.,23 the total cost of hip fractures was estimated by identifying all hip fracture–related health resource utilization in the U.S. and multiplying by the corresponding cost per unit of service. By contrast, OTA estimated the total costs of hip fractures by multiplying average inpatient hospital, nursing home, and outpatient expenditures per patient by the total number of hip fracture patients in the U.S. The criterion used to identify hip fracture–related medical services vary from study to study. For example, this study used both primary and secondary diagnoses to identify hip fracture hospitalizations, whereas Praemer et al.24 used primary diagnoses only.
The specific inpatient and outpatient medical services that are included in each study vary as well. For example, outpatient physical therapy services were included in this analysis but not in the OTA study.8 Finally, the cost perspective and sources of data vary from study to study. The unit prices employed by OTA8 were based primarily on Medicare allowed charges, whereas Praemer et al.24 and Holbrook et al.23 used both submitted and allowed charges. The cost data used in this study, obtained from the National Medical Expenditure Study, represent “…either allowed charges when total submitted charges are reduced to the amounts allowed by third-party payers or total charges if there was no such reduction.”47 The variation in unit prices is demonstrated by examining, for example, the average cost per inpatient day (inflated to 1995 dollars) used in each study, which ranges from $985 per day (Holbrook et al.23) to $1044 per day (OTA)8 to $1620 per day (Praemer et al.24) and $1851 per day (this study). Further compounding the difference in per patient costs among these four studies is the change in medical practice patterns in this country during the last 15 years. For example, the average length of an inpatient hospital stay for hip fracture has dramatically declined from 18.6 days in 1982 to 12.3 days in 1991.48,49
The expenditure estimates presented in this study were based on the most current and reliable data available and the osteoporosis attribution probabilities developed by an Expert Panel specifically for this study.34 Nevertheless, a number of limitations should be noted. All encounters with a medical diagnosis indicating treatment of an osteoporotic fracture were identified and extracted from each data source. Thus, the estimates will be affected to the extent that fractures are under- or over-represented as a primary and secondary diagnosis. For example, up to 6% of hip fracture cases may go unreported in Medicare hospitalization data.10 Despite the fact that the World Health Organization's definition of osteoporosis is based on bone density levels rather than fractures,50 our study did not include costs associated with persons diagnosed with osteoporosis during 1995 who did not have a fracture. This is because of the difficulty of using secondary data to identify persons being screened for osteoporosis. Additionally, persons with undiagnosed osteoporosis were not included because undiagnosed osteoporosis has no concurrent cost and, therefore, cannot be included in our prevalence-based estimates. However, both of these groups have potential future costs related to the prevention and treatment of osteoporotic fractures.
Furthermore, our analysis did not examine the costs associated with nonhealth sector goods and services, such as transportation to medical care, special diets, patient education, and alterations to a patient's home, nor the indirect costs associated with lost productivity of affected patients and caregivers or premature mortality. These direct nonmedical and indirect costs were not included because of the unavailability of data specific to persons with osteoporosis who have an incident fracture. Such costs may be substantial, however. Praemer et al.24 estimated direct nonmedical costs, indirect morbidity, and indirect mortality costs associated with all hip fractures in 1988 at 1.2, 1.4, and $260 million, respectively. In addition, we have not included support costs related to government expenditures for osteoporosis-related research or private monies spent in the development of new technologies to detect, prevent, and treat osteoporosis. By excluding these costs, our results must be considered conservative.
In conclusion, our study provides further evidence that osteoporosis is an important public health problem that contributes to a significant proportion of fractures and associated health expenditures in this country. The total 1995 health expenditures for osteoporotic fractures amount to 1.5% of National Personal Health Care Expenditures and 2.4% of National Hospital Care Expenditures.51 The results of this study suggest that, once a fracture occurs, substantial health care expenditures are incurred. Therefore, evaluation of screening programs and interventions used to prevent osteoporotic fractures are warranted. Our study also suggests that, despite current perceptions, the impact of osteoporosis is not limited to hip fractures among the oldest white women. The finding that nearly two-thirds of total expenditures associated with osteoporotic fractures are attributable to the population aged 45–84 years suggests that preventive educational campaigns should be initiated at midlife or earlier. If such efforts resulted in the prevention or amelioration of osteoporosis, a substantial reduction in morbidity and costs would be realized since 91% of total osteoporotic medical expenditures are associated with short- and long-term institutional care. Improved methods of diagnosis and treatment for osteoporosis are important since the number of osteoporotic fractures is expected to increase dramatically with the rising number of elderly individuals in the population. According to one projection, demographic changes alone could lead to an increase in the number of hip fractures annually to 840,000 by the year 2040.52 Finally, the finding that 7% of the total direct medical expenditures were attributed to minorities and 20% to men provides additional empirical support for the contention that osteoporosis is a condition that affects the entire aging population.53
This study was supported in part by a grant from Merck & Co, West Point, Pennsylvania, U.S.A.