Potential Impact of Osteoporosis Treatment on Hip Fracture Trends

The efficacy of osteoporosis treatment with antiresorptive agents has been shown in a number of large and well-designed clinical trials, but the effectiveness of these interventions to reduce the burden of fractures in the general population is less certain. Various studies have shown that relatively few people are screened for osteoporosis1 and that many of those who are evaluated subsequently discontinue treatment after only a few months.2, 3 Moreover, the majority of patients who sustain an osteoporotic fracture are not treated at all,4 even though they are at greatly increased risk of another one.5 In this context, Jaglal et al.,6 in this issue of the Journal, suggest that aggressive diagnosis and treatment of osteoporosis in Canada is having a very substantial effect in lowering the incidence of osteoporotic fractures and that predictions of rising numbers of hip fractures in the future are, therefore, misleading. They base this view on the observation that the estimated age-adjusted incidence of hip fractures in Canadian women residing in Ontario fell by about 20% between 1992 and 2005, with a similar decline in men, while wrist fracture incidence decreased by a lesser amount. These changes were attributed to a marked concurrent increase in the number of people filling at least one prescription for an antiresorptive agent (mostly etidronate, but excluding hormone therapy), from about 13,000 in 1996 to nearly 226,000 in 2003. This, in turn, was presumably in response to a 10-fold increase in BMD testing from about 37,000 DXA tests in 1992 to just over 404,000 in 2001. On the basis of these trends, the authors conclude that greater diagnosis and treatment of osteoporosis has reduced the number of hip fractures below previously projected figures, even in the face of an increased number of elderly Canadians.6 These results are promising, and they raise two main questions. First, do these data prove that the hip fracture projections were "wrong," as indicated in the title of their paper? Second, are the reported trends in BMD testing and prescribing antiresorptives the cause of the fracture reduction that was seen? On reflection, it should be clear that hip fracture projections do not represent a prophecy but rather a warning of what could happen later unless more serious attempts are made to control the main determinants of hip fracture risk (i.e., skeletal fragility and falls). Because comprehensive interventions to prevent falls are challenging and of somewhat limited effectiveness,7 the focus of most fracture prophylaxis schemes centers on preventing bone loss.8 Indeed, when taken at face value, the results of Jaglal et al.6 show the value of the hip fracture projections that were made earlier for the Canadian population (i.e., they stimulated a greater effort to manage osteoporosis, and the burden of hip fractures was reduced as a consequence). It does not necessarily follow, however, that this is a generalizable result. Previously published hip fracture projections were based on clearly stated assumptions about secular trends in fracture incidence, combined with estimates of future population growth. Because hip fracture rates rise exponentially with age,9 it is growth in the elderly population that is most germane. In the United States in the 14 years between 1992 and 2005, the period covered by Jaglal et al.,6 the population ⩾65 years of age grew by about 14%; however, it is projected to increase by 44% in the next 14-year period (2006-2019), as the wave of post-World War II "baby boomers" moves into this age group. Even larger increases are expected in Canada, where the number of men and women ⩾65 years of age is projected to increase by 62% and 51%, respectively, between 2005 and 2020.10 These projections are robust because all individuals who will be ⩾65 years of age in 2020 have already been born. Thus, the benefits of reduced hip fracture rates could yet be overwhelmed by a larger number of elderly residents. In other parts of the world, this is virtually guaranteed to be so. The population ⩾65 years of age in Asia, for example, is expected to increase 2.3-fold, from 921 million to 2.2 billion, between 1990 and 2025.11 Moreover, there is no indication that the use of effective osteoporosis interventions is increasing dramatically in those regions where growth in the elderly population is most rapid. The second important question is whether the decrease in fracture incidence can be shown to be caused by the osteoporosis interventions described. This is problematic for two reasons that relate to the pattern of secular trends and the performance characteristics of BMD tests. First, the assessment of secular trends in hip fracture incidence is complex. For example, age-adjusted hip fracture incidence rates in women from Rochester, MN, began to rise rapidly after 1930, but the rapid increase among European women seems to have begun about two decades later.12 Some ascribe the increase in hip fracture incidence to a greater prevalence of frailty in the population, with a corresponding increase in the likelihood of falling,13 and there is little reason to believe that frailty or falling would be counteracted by the use of antiresorptive agents.14 Indeed, it has been suggested that the age-related increase in falls and other factors is substantially more important than the age-related decrease in hip BMD in accounting for the rise in hip fracture incidence rates with age.15 Conversely, no convincing explanation has emerged for the observation that hip fracture incidence in some of these same regions subsequently stabilized or fell.16 Thus, it is important to note that the declining hip fracture incidence rates seen in Rochester women since 1950 and in Rochester men since 198017 parallel those now reported for Ontario, a similar climatologic region, yet predate the introduction of osteoporosis practice guidelines by decades. As the authors point out, their study is an ecologic study and, as such, is potentially subject to the "ecologic fallacy," (i.e., the presumption that changes seen at the population level reflect those changes actually taking place in individuals). Thus, a remarkable 23% of Ontario women filled a prescription for etidronate, alendronate, risedronate, or raloxifene in 2003, although it is not clear the extent to which these new prescriptions may have reflected treatment changes among women whose bone loss was previously controlled by long-term hormone therapy.18 However, the rapid decline in hip fracture incidence began in 1997 and, in the preceding year, <2% of women were so treated. Acute reductions in fracture risk can be seen in elderly individuals, even those who already have experienced considerable bone loss, because antiresorptive treatment can lower fracture risk substantially by virtue of slowing bone turnover.19 This is not as obvious an explanation for the decline in hip fracture incidence in men, who were less intensively treated. Confidence in the conclusions of Jaglal et al.6 would be increased by showing some concordance between the rates of intervention in specific age and sex subgroups and the contemporary reductions in hip fracture risk seen in those same groups. The second issue is that the decrease in hip fracture risk seems larger than can be explained by BMD testing. The expected decrease in hip fracture rates by BMD testing can be estimated from the use of BMD tests and their performance characteristics. In 2001, there were ∼1.8 million women ⩾50 years of age in Ontario, suggesting that just >20% of the postmenopausal women received a BMD test that year. Assuming that all BMD testing represented DXA at the hip and that hip fracture risk would increase about 2.5-fold for each SD decrease in BMD, the sensitivity of the test over a 10-year period would be ∼16% at 65 years of age.20 Further assuming 100% compliance with a 5-year treatment, roughly 20% of hip fractures would be averted among the screened patients. However, because the majority of fractures would occur in the unscreened population in this example, the overall impact would be to reduce hip fractures in the general community by only about 5%. Even assuming that the increase in BMD testing and antiresorptive therapy had influenced the incidence of distal forearm fractures, 44% of which occur in women between 55 and 74 years of age,21 the age range when bone densitometry was most frequently undertaken in Ontario, the impact on hip fracture risk might still be limited. It is true, as the authors point out, that patients with forearm fractures are at increased risk of a subsequent hip fracture.5 However, that hip fracture occurs, on average, 14 years after the forearm fracture.22 Therefore, the full benefit of treating middle-aged women for osteoporosis in recent years may not become evident for decades. Indeed, the long-term benefits are somewhat uncertain because most patients are unlikely to remain on therapy for such an extended period of time. Moreover, antifracture benefits long after cessation of treatment have yet to be quantified, although there do seem to be persistent positive effects of bisphosphonate therapy with respect to BMD and bone turnover.23 In conclusion, the observations of Jaglal et al.6 are extremely encouraging despite some concern about causal relationships that should not discourage efforts to optimize the management of osteoporosis. A direct demonstration that enhanced osteoporosis management actually reduces fractures would confound critics who contend that the clinical "high-risk" strategy (i.e., the active identification of individuals at greatest risk of fracture for treatment with a potent agent) is unlikely to have any real impact at the population level compared with a public health strategy of lowering community risk generally (i.e., the use of inexpensive interventions population-wide, without assessment of individual risk, such as vitamin D fortification of dairy products).24 There are recent indications that clinical attention to osteoporosis management is increasing25 and that specific programs of selective screening and aggressive treatment to prevent bone loss may really be able to reduce population fracture rates over the short-term.26, 27 Additional empirical evidence with regard to the effectiveness of osteoporosis interventions in the community would complement the extensive data already available on the efficacy of these interventions in the setting of randomized controlled trials in selected patient populations. This project was supported in part by National Institute on Aging Research Grant AG 04875.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.