By Heidi D. Nelson, M.D., M.P.H. a,b; Mark Helfand, M.D., M.P.H. a,b; Steven H. Woolf, M.D., M.P.H. c; Janet D. Allan, P.H.D., R.N. d.
Address correspondence to: Heidi D. Nelson M.D., M.P.H., Oregon Health & Science University, Mail code BICC 504, 3181 SW Sam Jackson Park Road, Portland, OR 97201; E-mail: nelsonh@ohsu.edu.
This article originally appeared in the Annals of Internal Medicine. Select for copyright and source information.
The summaries of the evidence briefly present evidence of effectiveness for preventive health services used in primary care clinical settings, including screening tests, counseling, and chemoprevention. They summarize the more detailed Systematic Evidence Reviews, which are used by the U.S. Preventive Services Task Force (USPSTF) to make recommendations.
Abstract
Introduction
Methods
Results
Discussion
Acknowledgments
References
Notes
Background: Although osteoporotic fractures present an enormous health burden, it is not clear whether screening to identify high-risk persons is appropriate.
Purpose: To examine evidence on the benefits and harms of screening postmenopausal women for osteoporosis.
Data Sources: MEDLINE® (1966 to May 2001), HealthSTAR (1975 to May 2001), and Cochrane databases; reference lists; and experts.
Study Selection: English-language abstracts that included original data about postmenopausal women and osteoporosis and addressed the effectiveness of risk factor assessment, bone density tests, or treatment were included.
Data Extraction: Selected information about patient population, interventions, clinical endpoints, and study design were extracted, and a set of criteria was applied to evaluate study quality.
Data Synthesis: No trials of the effectiveness of screening have been published. Instruments developed to assess clinical risk factors for low bone density or fractures have moderate-to-high sensitivity and low specificity. Among different bone density tests measured at various sites, bone density measured at the femoral neck by dual-energy x-ray absorptiometry is the best predictor of hip fracture. Women with low bone density have approximately a 40 percent to 50 percent reduction in fracture risk when treated with bisphosphonates.
Conclusions: Population screening would be based on evidence that the risk for osteoporosis and fractures increases with age, that the short-term risk for fracture can be estimated by bone density tests and risk factors, and that fracture risk can be reduced with treatment. The role of risk factor assessment and different bone density techniques, frequency of screening, and identification of subgroups for which screening is most effective remain unclear.
Half of all postmenopausal women will have an osteoporosis-related fracture during their lives, including one-quarter who will develop a vertebral deformity (1) and 15 percent who will suffer a hip fracture (2). Hip fractures are associated with high mortality rates and loss of independence (3,4). Although many vertebral fractures are detected only incidentally on radiography, some cause severe pain, leading to 150,000 hospital admissions per year in persons older than 65 years of age, 161,000 physician office visits, and more than 5 million days of restricted activity in those 45 years of age or older (5).
Low bone density has been used to predict risk for fractures as well as to diagnose osteoporosis. Osteoporosis has been defined as "a systemic skeletal disease characterized by low bone mass and microarchitectural deterioration of bone tissue, leading to enhanced bone fragility and a consequent increase in fracture risk" (6,7). This definition emphasizes that, in addition to bone mass, the structure of bone is also an important factor in the pathogenesis of fractures.
A World Health Organization working group (8) proposed that osteoporosis should be diagnosed in epidemiologic studies when bone mineral density is 2.5 standard deviations (SDs) or more below the mean for healthy young adult women at the spine, hip, or wrist (corresponding to a T-score of <-2.5) or when patients have a history of an atraumatic fracture (9). By the World Health Organization definition, up to 70 percent of women older than 80 years of age have osteoporosis (10). Age is also an important factor in the relationship between bone density and the absolute risk for fracture. Older women have a much higher fracture rate than younger women with the same bone density because of increasing risk from other factors, such as bone quality and tendency to fall (11).
Despite the high prevalence of osteoporosis and the effect of fractures on mortality, independence, and quality of life, it is not clear whether it is appropriate to screen asymptomatic postmenopausal women. Recent systematic reviews and guidelines disagree about which women should be screened and when (12-20). This disagreement reflects, in part, gaps in the evidence. For example, most guidelines recommend using risk factors to select patients for bone density testing, but because of inadequate data there is no consensus on which risk factors to use.
As part of the U.S. Preventive Services Task Force update of its 1996 recommendation (21), we examined evidence on the benefits and harms of screening postmenopausal women for osteoporosis. Specifically, we addressed the role of risk factors in identifying high-risk women, techniques of bone density testing to identify fracture risks, effectiveness of treatment in reducing fracture risk, and harms of screening and treatment.
Additional methods used for this review, including determination of the quality of studies (22), are detailed in Appendix Tables 1, 2, and 3 and in a separate report (23). The analytic framework and key questions are detailed in the Appendix Figure 1 (8 KB). Relevant studies were identified from multiple searches of MEDLINE® (1966 to May 2001), HealthSTAR (1975 to May 2001), and Cochrane databases; reference lists of systematic reviews; and experts. We also sent letters to manufacturers of bone density devices requesting additional information about the performance of their instruments, but we received no new data.
Two reviewers read each abstract to determine its eligibility. We included English-language abstracts that included original data about postmenopausal women and osteoporosis and addressed screening, or the effectiveness of risk factor assessment, bone density testing, or treatment. We considered screening to be the process of assessing postmenopausal women without known osteoporosis for risk of osteoporotic fractures by identification of risk factors, including low bone density. Postmenopausal women were those who had experienced surgical or natural menopause, regardless of age. Women with pre-existing atraumatic fractures were not considered in the screening population because they had confirmed osteoporosis according to the World Health Organization definition.
For studies of prediction, we selected articles that reported the relationship between risk factor assessment methods or bone density tests and bone density, bone loss, or fractures. We reviewed studies of medications used for treatment, and present results for bisphosphonates. We focused on randomized controlled trials of therapies reporting radiographically verified, nontraumatic fracture outcomes, because fractures are a stronger measure of effectiveness than bone density.
We excluded studies of primary prevention of osteoporosis, such as the role of nutrition, calcium consumption, and physical activity. We also excluded secondary causes of osteoporosis, such as corticosteroid use and certain chronic diseases, and studies that did not provide sufficient information to determine the method for selecting patients and for analyzing data. Investigators read the full-text versions of the retrieved papers and re-applied the initial eligibility criteria.
To assess the internal validity of individual studies, we applied a set of criteria developed by the current U.S. Preventive Services Task Force (Appendix Table 1) (22).
In this paper, we highlight studies that are applicable to current practice standards, have high quality internal validity ratings, and are most generalizable to the U.S. population of postmenopausal women under consideration for screening. We created an outcomes table to summarize the number of hip and vertebral fractures prevented based on age-specific prevalence rates and treatment effects obtained from results of the reviewed studies. We conducted a sensitivity analysis to determine the influence of risk factors on the number needed to screen.
This research was funded by the Agency for Healthcare Research and Quality (AHRQ) under a contract to support the work of the U.S. Preventive Services Task Force. Agency staff and Task Force members participated in the initial design of the study and reviewed interim analyses and the final manuscript. Additional reports were distributed for review to content experts and revised accordingly before preparation of this manuscript (23,24).
Hundreds of studies report associations between risk factors and low bone density and fractures in postmenopausal women (24). The most comprehensive study of risk factors in a U.S. population is the Study of Osteoporotic Fractures (SOF), a good-quality prospective study of 9,516 women 65 years of age and older (25). In this study, 14 clinical risk factors were identified as significant predictors of hip fracture in multivariable models (age; maternal hip fracture; no weight gain; height; poor self-rated health; hyperthyroidism; current use of benzodiazepines, anticonvulsants, or caffeine; not walking for exercise; lack of ambulation; inability to rise from a chair; poor scores on two measures of vision; high resting pulse; and any fracture since 50 years of age). The relative risk for hip fracture per decrease of 1 SD in calcaneal bone density was 1.6 (95 percent CI, 1.3-1.9). This was comparable to the magnitude of the relative risks of most of the other significant predictors in the model, which ranged from 1.2 to 2.0. Women with at least 5 of the 14 risk factors had increased rates of hip fractures compared with those who had 0 to 2 risk factors at all levels of calcaneal bone density.
To determine which risk factors could be important in women younger than 65 years of age, we reviewed eight observational studies of risk factors and fractures of various types conducted in populations in which at least 50 percent of participants were younger than 65 years of age. Table 1 lists risk factors that were statistically significant predictors for fractures in multivariable models (26-33). These results could not be quantitatively combined because risk factors were defined differently in each study.
Results of risk factor studies have been used to assess risk in individuals. We identified 10 cross-sectional studies that described methods of determining risk for low bone density for individual women based on selected clinical risk factors (Table 2) (34-43). The most common methodologic limitations of these studies are lack of validation and lack of generalizability because of small numbers of patients or nonrepresentative patients. We also identified eight studies of clinical risk factors to determine fracture risk (Table 2) (44-51). None of these studies received a good rating for internal validity. Four studies evaluated hip fracture outcomes, two evaluated vertebral fractures, and two evaluated all types of fractures. These studies described the association of risk factors with fractures known to have occurred already (four case-control studies), or how well they would predict fractures in the future (four prospective cohort studies).
A recent study compared the performance of 5 clinical decision rules for bone density testing among 2,365 postmenopausal women 45 years of age or older who were enrolled in a community-based study of osteoporosis in Canada (52). These rules included guidelines from:
None of the decision rules had good discriminant performance. In this study, SCORE and ORAI had the highest area under the receiver-operating characteristic curves (0.80 and 0.79, respectively [sensitivity, 99.6 percent and 97.5 percent, and specificity, 17.9 percent and 27.8 percent respectively]). Details of how to use SCORE and ORAI are given in Table 2).
Several technologies are available to measure bone density (55-58), although correlations among different bone density devices are low (0.35 to 0.60) (59-79). Dual-energy x-ray absorptiometry (DXA) is considered the gold standard because it is the most extensively validated test against fracture outcomes. When used in the same patients, DXA machines from different manufacturers differ in the proportion of patients diagnosed to have osteoporosis by 6 to 15 percent (80-85). Published studies consistently show that the probability of receiving a diagnosis of osteoporosis depends on the choice of test and site (86-90). One analytical study, for example, found that 6 percent of women older than 60 years of age would receive a diagnosis of osteoporosis if DXA of the total hip were used as the only test, compared with 14 percent for DXA of the lumbar spine, 3 percent with quantitative ultrasonography, and 50 percent with quantitative computed tomography (87).
The likelihood of receiving a diagnosis of osteoporosis also depends on the number of sites tested. Testing in the forearm, hip, spine, or heel generally identifies different groups of patients. For example, a physician cannot definitively say that a patient does not have osteoporosis on the basis of a forearm test alone. Conversely, although test results at any site are associated to some degree with fractures at other sites, a physician may not be able to assess whether a patient with a low T-score on a hand or forearm test has substantial bone loss at other sites.
A meta-analysis assessed 23 publications from 11 separate prospective cohort studies published before 1996 (91). Nearly all of the data were from women in their late 60s or older. No studies of ultrasonography were included. The meta-analysis indicated that DXA at the femoral neck predicted hip fracture better than measurements at other sites, and was comparable to forearm measurements for predicting fractures at other sites (92-94). For bone density measurements at the femoral neck, the pooled relative risk per decrease of one SD in bone density was 2.6 (CI, 2.0-3.5). In direct comparisons, heel ultrasonography was slightly worse than but comparable to DXA of the hip in women older than 65 years of age (Table 3) (92,94-100). For both tests, a result in the osteoporotic range is associated with an increased short-term probability of hip fracture. No data compare DXA and ultrasonography for prediction of fracture in women younger than 65 years of age.
The National Osteoporosis Risk Assessment study (30) recently evaluated the performance of peripheral densitometry in predicting fractures. This prospective study of ambulatory postmenopausal women 50 years of age or older with no previous osteoporosis diagnoses recruited 200,160 participants from 4,236 primary care practices in 34 U.S. states. Women received baseline T-scores by measuring bone density at the heel (single-energy x-ray absorptiometry or quantitative ultrasonography), forearm (peripheral DXA) or finger (peripheral DXA). After 12 months of followup, women with T-scores less than or equal to -2.5 had an adjusted risk for all types of fractures that was 2.74 (CI, 2.40-3.13) times higher than women with normal baseline bone density. Results varied by type of test and site; those identified as osteoporotic by DXA had higher fracture rates. Tests were not compared with DXA of the femoral neck, and the study did not describe how tests performed by age group or risk category.
The U.S. Food and Drug Administration has approved hormone replacement therapy, bisphosphonates, raloxifene, and calcitonin for osteoporosis prevention and/or treatment. Our review of estrogen and selective estrogen receptor modulators is presented elsewhere (101). A recent meta-analysis (102) of 11 randomized trials (103-113) enrolling 12,855 women found that at least 5 mg of alendronate per day reduced vertebral fractures in 8 trials (relative risk, 0.52 [CI, 0.43-0.65]). Alendronate also reduced:
These trials included followup data ranging from 1 to 4 years; effect sizes for longer periods of use are not known. We evaluated data from these trials to determine whether women who have a similar overall risk for fracture but different bone densities derive a similar benefit from treatment. This question is clinically important because accepted criteria for initiating treatment are lacking.
The Fracture Intervention Trial (FIT) of alendronate was conducted with two groups of participants and provides some information about levels of risk. One group (FIT-I) included a higher-risk sample of 2,027 women who had T-scores of -1.6 or lower and pre-existing vertebral fractures (104). The 3-year risk for hip fracture was 2.2 percent in the placebo group and 1.1 percent in the alendronate group (relative hazard, 0.49 [CI, 0.23-0.99]), and the 3-year risk for any clinical fracture was 18.2 percent in the placebo group and 13.6 percent in the alendronate group (relative hazard, 0.72 [CI, 0.58-0.90]). A second study from FIT (FIT-II) included a lower-risk sample of 4,432 women who also had T-scores of -1.6 or lower, but did not have pre-existing vertebral fractures (114). The 4-year incidences of hip fracture (1.1 percent) and any clinical fractures (14.1 percent) in the placebo group were lower than those observed in the FIT-I placebo group. In FIT-II, only the subgroup of treated patients who had T-scores lower than -2.5 (n = 1,627) had a significant risk reduction for all clinical fractures, from 19.6 percent to 13.1 percent (relative risk, 0.64 [CI, 0.50-0.82]). No reduction in risk for fractures was seen in patients who had T-scores between -1.6 and -2.5.
The results from FIT suggest that women with more risk factors for fracture relating to bone structure and integrity, such as age, very low bone density, or pre-existing vertebral fractures, derive the greatest absolute benefit from treatment. However, FIT did not examine other nonskeletal risk factors, such as psychomotor impairment, poor gait, and other factors that increase the risk for falling. The effect of some of these risk factors on the benefit of treatment was examined in a randomized trial of another bisphosphonate, risedronate (115). Risedronate had no effect on hip fracture rates among women 80 years of age or older who had one or more risk factors for falls but who did not necessarily have low bone density. In the same report, in women 70 to 79 years of age with severe osteoporosis (T score < -3), risedronate reduced hip fractures by 40 percent (relative risk, 0.6 [CI, 0.4-0.9]; number needed to treat for benefit, 77).
Trial results are not applicable to a screening program unless the trials included patients who would be identified by screening the general population. We examined recruitment and eligibility characteristics of 10 of the 11 published randomized trials of alendronate to assess whether selection biases or other biases might have affected their generalizability (Table 4) (103-112). Overall, the trials included relatively healthy women with low bone density who were not using estrogen. Except for participants in two trials involving women who had recently gone through menopause and were not osteoporotic, most participants were older than 65 years of age.
The FIT-II is the largest study and provided the most detailed description of recruitment and results (107). In FIT-II, researchers recruited the sample of 4,432 women by mailing a query to more than 1 million women selected from the general population in 11 cities. Women who had medical problems (for example, dyspepsia) or who used estrogen were excluded. Fifty-four thousand women (approximately 5.4 percent) responded by telephone; of these, 26,137 (52 percent) had a screening visit. A higher than expected proportion of these (65 percent) had sufficiently low bone density to enroll in the study. Of this 65 percent, 57 percent were classified as "ineligible, did not wish to continue, or screened after recruitment to this arm." It is not clear from this description how many patients did not meet the eligibility criteria. In addition, an unspecified number of patients (up to 28,000) were found to be ineligible at the initial stage of recruitment. The demographic characteristics of eligible and screened but excluded participants were not reported. None of the other randomized trials disclosed any details of how their samples were recruited or how many respondents were found to be ineligible before randomization.
In other clinical areas, the results of industry-sponsored trials were significantly more favorable to newer therapies than trials funded by nonprofit organizations (116,117). Because all 11 trials of alendronate were funded wholly or in part by the manufacturer, we were unable to assess the influence of sponsorship on effect size. If effectiveness of treatments is less than estimated in these trials, the efficiency of screening to identify candidates for treatment will be reduced and the number needed to screen for benefit will increase.
Several potential harms are associated with screening and treatment. A test result indicating osteoporosis could produce anxiety and perceived vulnerability (118) that may be unwarranted. On a quality-of-life questionnaire, women with osteoporosis voiced significantly more fears than women who had normal bone density (119). Some women may be falsely reassured if abnormal results from last year's DXA appear "improved" on this year's normal calcaneal ultrasonogram. The potential time, effort, expense, and radiation exposure of repeated scans over a lifetime have not yet been determined.
Potential harms may also arise from inaccuracies and misinterpretations of bone density tests. The variation among techniques, along with the lack of methods to integrate bone density results with clinical predictors, makes it difficult for clinicians to provide accurate information to patients about test results. In one study, physicians found densitometry reports confusing, and were not confident that their interpretations of T-scores were accurate (120). False-positive results could lead to inappropriate treatment, and false-negative results could lead to missed treatment opportunities.
Harms of treatment depend on the medication used. Overall, gastrointestinal side effects occur in approximately 25 percent of patients taking alendronate, but in controlled trials these rates were usually not significantly higher than those for placebo. High rates of serious gastrointestinal side effects have been observed among Medicare enrollees taking alendronate (121). The long-term adverse effects of alendronate are unknown.
Costs of screening vary with technique, and average 2000 Medicare reimbursement rates were $133 for DXA and $34 for ultrasonography (122). Abnormal results on ultrasonography may require a confirmatory DXA before treatment because clinical trials are based on entry criteria using DXA. Most women would require repeated tests over several years before receiving a diagnosis of osteoporosis and leaving the screening pool. Treatment costs also vary; alendronate currently costs approximately $3 per daily dose.
To estimate the effect of screening 10,000 postmenopausal women for osteoporosis on reducing hip and vertebral fractures, we created an outcomes table based on assumptions from the reviewed studies (Table 5). These estimates include age-specific prevalence rates expressed in 5-year age intervals (123) and treatment effects based on trial results (risk reduction, 37 percent for hip fracture and 50 percent for vertebral fracture) (102,104,115,124). We estimated an adherence rate of 70 percent based on reports of adherence and side effects from treatment trials, assuming less optimal compliance in the general population.
When the assumptions in Table 5 are used, if 10,000 women 65 to 69 years of age underwent bone densitometry (DXA of the femoral neck), 1,200 would be identified as high-risk (T-score = -2.5). If these women were offered treatment that resulted in a 37 percent reduction in hip fracture risk, and a 50 percent reduction in vertebral fracture risk, and 70 percent adhered to therapy, then 14 hip fractures and 40 vertebral fractures would be prevented over a 5-year period. The number of women in this age group needed to screen to prevent one hip fracture in 5 years would be 731, and the number of women with low bone density needed to treat for benefit would be 88. The number needed to screen to prevent one vertebral fracture would be 248, and the number needed to treat for benefit would be 30. Treatment has significant costs and potential harms; when the number-needed-to-screen for benefit is high, the balance of benefits and harms may become unfavorable. These numbers become more favorable in older persons because the prevalence of osteoporosis increases steadily with age.
There is interest in whether risk assessment can be used to select patients for bone densitometry, which is costly. Our literature review indicated that the prevalence of osteoporosis, the predictability of densitometry, and the effectiveness of treatment might be lower for younger than for older postmenopausal women. To determine whether it is useful to consider clinical risk factors when screening younger postmenopausal women, we also included risk estimates for clinical risk factors in a sensitivity analysis. Our review of observational studies with younger postmenopausal women indicated that three consistent predictors of fracture are increasing age, low weight or body mass index, and nonuse of hormone replacement therapy (defined by current use, ever use, or certain durations of use). These three variables are also used in ORAI to identify women with low bone density (43) and were the variables most strongly associated with low bone density in a study enrolling mostly younger postmenopausal women in the United States (125). On the basis of these studies, we estimated that one of these risk factors increases the probability of having osteoporosis by up to 100 percent and increases the risk for fracture by 70 percent (relative risk, 1.7).
For the younger age groups, the presence of clinical risk factors influences the outcomes. For example, only five hip fractures are prevented over 5 years when all women 60 to 64 years of age are screened; however, nine hip fractures are prevented if women have a factor that increases fracture risk by 70 percent. For women 60 to 64 years of age who have such a risk factor, the number needed to screen is 1,092 and the number needed to treat for benefit is 72 to prevent 1 hip fracture. These numbers approach those of women 65 to 69 years of age (Figure 1).
Although many studies have been published about osteoporosis in postmenopausal women, no trials have evaluated the effectiveness of screening; therefore, no direct evidence that screening improves outcomes is available. Instruments developed to assess clinical risk factors for low bone density or fractures generally have moderate to high sensitivity and low specificity. Many have not been validated, and none have been widely tested in a practice setting. Among different bone density tests measured at various sites, bone density measured by DXA at the femoral neck is the best predictor of hip fracture and is comparable to forearm measurements for predicting fractures at other sites. Heel ultrasonography and other peripheral bone density tests, however, can also predict short-term fracture risk. Bisphosphonates decrease fracture risk by approximately 40 to 50 percent in women with low bone density.
Support for population screening would be based on evidence that the prevalences of osteoporosis and fractures increase with age, that the short-term risk for fracture can be estimated by bone density tests and risk factors, and that the fracture risk among women with low bone density can be significantly reduced with treatment. We applied these data to generate an outcomes table of screening strategies that provide estimates of the numbers of women needed to screen and treat to prevent fractures. Age-based screening is supported by prevalence data, that is, the number needed to screen to prevent fractures decreases sharply as age and prevalence increase. Use of risk factors, particularly increasing age, low weight, and nonuse of estrogen replacement, to screen younger women may identify additional high-risk women and provide absolute benefit similar that yielded by screening older women without risk factors. These findings relate to screening asymptomatic women only and do not apply to women considered for testing because of pre-existing or new fractures or the presence of secondary causes of osteoporosis.
Our approach has several limitations, however, and results from a well-designed trial of screening strategies should supersede our estimations which are based on indirect evidence. The estimates in the outcomes table are limited by assumptions that are arguable or highly variable by patient and setting. Our assumptions of treatment effect and adherence are especially optimistic and reflect results of clinical trials, not clinical practice. We chose a 5-year time horizon based on the short-term predictability of bone density tests as well as on results of short-term treatment trials. Long-term outcomes may provide a more accurate estimate of benefits. Also, we cannot exclude the possibility that harms outweigh benefits, particularly since the long-term effects of bisphosphonates are not yet known.
The evidence on which we based our conclusions is also limited. Overall, evidence is stronger for women older than 65 years of age than for younger women because more research has been done in older age groups. Bone loss in the perimenopausal and early postmenopausal years is important to long-term bone health, but few published studies address screening and treatment for younger postmenopausal women. Fracture risk is determined not only by bone density, but also by bone characteristics that are difficult to measure in a clinical setting, such as bone structure and morphologic characteristics. No bone density studies or treatment trials include large numbers of non-white women, and it may be difficult to provide ethnicity-specific screening recommendations in the absence of more evidence.
The role of clinical risk factors is still unclear. Although many risk factors are associated with osteoporosis and fractures, how to use them to select women to test or treat is uncertain. The risk factors identified by our literature review and used in the outcomes table are only best estimates. Other risk factors may prove to be equally predictive when used for screening purposes. Further validation of existing risk assessment instruments or development of new ones would be useful. Few studies have evaluated the effect of altering modifiable risk factors, such as smoking cessation, strength and balance training, and visual correction. These interventions may prove to be as effective as drug therapy in preventing fractures, and may also be important effect modifiers that would alter the effectiveness of treatments.
Peripheral bone density tests have not been extensively studied for screening. Results from National Osteoporosis Risk Assessment Study (30) indicate that peripheral tests can predict short-term fracture rates in a primary care population that would be targeted for screening. Most treatment trials use DXA of the hip as an entry criterion, and results may not apply to women whose diagnosis is determined by other tests. A sequential approach, in which women with low values on a peripheral test are subsequently tested by DXA of the hip to determine treatment needs, may be useful, although this approach has not been evaluated. Further research is needed to define the appropriate use of these technologies.
How frequently to screen has also not been specifically studied, but data are needed to determine optimal screening intervals. Estimations can be made based on the age-specific prevalence of osteoporosis and the precision of bone density tests. Less frequent testing for younger postmenopausal women when prevalence is lower (for example, 5-year intervals) and more frequent testing for older women (for example, 2-year intervals) might be reasonable, but further research is needed. Screening intervals of less than 2 years seem unwarranted because the precision error of densitometry would likely exceed the estimated bone loss in such a brief period (126). After a woman is screened and determined to have osteoporosis, future screening with bone density testing would be unnecessary.
Osteoporotic fractures present an enormous health burden on an expanding elderly population. Further research to more accurately determine the benefits and harms of screening is of paramount importance.