Volume 58, Issue 4 p. 738-745
PROGRESS IN GERIATRICS
Free Access

Prevention of Hip Fractures in Long-Term Care: Relevance of Community-Derived Data

Richard G. Crilly MD

Richard G. Crilly MD

From the Specialized Geriatric Services, St. Joseph's Health Care London, London, Ontario, Canada

Aging, Rehabilitation, and Geriatric Care Research Centre, Lawson Health Research Institute, London, Ontario, Canada

Division of Geriatric Medicine, Faculty of Medicine

Search for more papers by this author
Loretta M. Hillier MA

Loretta M. Hillier MA

From the Specialized Geriatric Services, St. Joseph's Health Care London, London, Ontario, Canada

Aging, Rehabilitation, and Geriatric Care Research Centre, Lawson Health Research Institute, London, Ontario, Canada

Search for more papers by this author
Michelle Mason BHSc

Michelle Mason BHSc

Health and Rehabilitation Sciences, Faculty of Health Sciences, University of Western Ontario, London, Ontario, Canada.

Search for more papers by this author
Iris Gutmanis PhD

Iris Gutmanis PhD

From the Specialized Geriatric Services, St. Joseph's Health Care London, London, Ontario, Canada

Aging, Rehabilitation, and Geriatric Care Research Centre, Lawson Health Research Institute, London, Ontario, Canada

Search for more papers by this author
Lizebeth Cox BHSc

Lizebeth Cox BHSc

Health and Rehabilitation Sciences, Faculty of Health Sciences, University of Western Ontario, London, Ontario, Canada.

Search for more papers by this author
First published: 01 April 2010
Citations: 19
Address correspondence to Dr. Richard Crilly, St. Joseph's Health Care London, 801 Commissioners Road East, London, ON, Canada N6C 5J1. E-mail: [email protected]

Abstract

Osteoporosis and falling are two major contributing factors to fractures in older persons; the relevant contribution of these may vary according to age, setting, and frailty. The purpose of this review was to examine the existing evidence on osteoporosis treatments to determine whether participants in clinical trials include or resemble the older and frailer adult population living in long-term care (LTC).

The trials (N=50) used to support major Canadian guidelines for osteoporosis treatment were reviewed because these are used to recommend treatment for all older adults, and several more-recent studies were added. Trials conducted specifically with participants living in LTC were also reviewed (N=6).

The majority of studies (96.0%) on osteoporosis treatments were conducted with community-dwelling participants, with many excluding participants resembling the LTC population. Mean ages ranged from 52 to 84, although for the majority of studies, the mean age was younger than 70. Similarly, 80.0% of studies conducted in LTC included only residents who were ambulatory, mobile, able to transfer independently, or not permanently bedridden. Mean ages in these studies ranged from 83 to 85.

These findings suggest that frail older adults, particularly the oldest and frailest adults in LTC, are neglected in clinical trials of osteoporosis fracture prevention. There is little evidence to support the application of community-based guidelines to the LTC population, and studies directly involving this population are needed. The role of age, frailty, and the mechanics of falls in hip fracture are discussed.

Osteoporosis is common in older people, and hip fracture is generally accepted as being the most significant manifestation of this problem. Prevention of hip fracture and other nonvertebral fractures is considered an important outcome in studies of pharmacological interventions for osteoporosis. Randomized controlled trials of osteoporosis medications to prevent fractures are generally conducted with selected populations, with specified admission and exclusion criteria. Beyond methodological requirements such as the need for blinding and randomization, there are the needs to obtain significant results within the constraints of time and cost. Attempts are therefore made to ensure that the study participants are likely to have a high enough event rate, be fairly adherent, attend testing, and live long enough to complete the study. In the case of trials of osteoporosis medication, this would require participants to be mobile and relatively young. The challenge is to conduct a study that meets these criteria but retains sufficient generalizability or external validity to be applicable beyond the study setting and the study population, for example, to older, frailer adults living in institutional settings, although it is acknowledged that, for the reasons outlined above, conducting trials in the appropriate setting, such as with the nursing home population, can be particularly difficult.

The purpose of this review was to examine the existing evidence on osteoporosis treatments to determine whether participants in clinical trials include those who resemble the older and frailer adult population living in long-term care (LTC) settings. To this end, the trials used to support major Canadian guidelines for osteoporosis treatment were reviewed because these are the guidelines used to recommend treatment for all older adults, regardless of age, frailty, functional status, and living situation.1,2 In addition to the treatments included in the Canadian guidelines, trials of zoledronic acid were reviewed, because this is a newer intervention not available at the time of the development of the Canadian guidelines. For comparative purposes, trials conducted specifically with participants living in LTC settings were also reviewed.

METHODS

The 2002 clinical practice guidelines for the management of osteoporosis and the 2006 Canadian Consensus Conference on Osteoporosis clinical practice update formed the basis for the current review.1,2 All of the studies used to support recommendations for the use of bisphosphonates, calcitonin, calcium and vitamin D, parathyroid hormone 1–34 (PTH), and hormone replacement therapy (HRT) were identified and reviewed for inclusion in this review. In addition, an independent search was conducted on zoledronic acid because studies on this bisphosphonate were not available when the guidelines were published. Studies were retained when the agent of interest was compared with placebo alone rather than in combination with other agents (but including those that involved vitamin D or calcium supplementation). Studies that did not examine fracture rates were also excluded. (Studies examining fracture rates as adverse events were retained for inclusion.) For each of the selected studies, information was collected on the number, sex, mean age, and standard deviation (SD) of participants in the intervention group(s) and the study population from which the participants were recruited (living situation, inclusion criteria). For studies in which there was more than one intervention group, as for example, studies that examined varying doses of the agent of interest, pooled mean ages and SDs were calculated. For consistency, when standard errors of the mean were presented, they were converted to SDs. One author (LMH) initially reviewed all of the studies to identify those that met the inclusion criteria; another author (RC) verified the selection of the studies.

From the 2002 clinical guidelines and 2006 update, a total of 153 studies that were used to support recommendations for the use of bisphosphonates, calcitonin, calcium and vitamin D, PTH, and HRT were identified (118 from the 2002 practice guidelines, 35 from the 2006 update). Two subsequent studies on zoledronic acid were identified. A total of 50 studies met the inclusion criteria for the current review (Figure 1): 48 studies identified from the 2002 and 2006 clinical guidelines and two additional studies on zoledronic acid.

Details are in the caption following the image

Selection of studies for review. *In addition, there were three meta-analyses that were not included in this review because the studies therein were already identified. Studies of the agent “alone” usually have vitamin D and calcium administered to both arms of the study. Plus one meta-analysis. §This includes eight articles that were identified in both the 2002 and 2006 guidelines and two additional studies on zoledronic acid. HRT=hormone replacement therapy; PTH=parathyroid hormone.

A search strategy for the review of the literature on intervention studies conducted with participants living in nursing home and LTC settings included the systematic search of several computerized bibliographic databases (MedLine, PubMed, Cochrane databases, CINAHL, Embase) using the following key words: LTC, nursing home, osteoporosis, vitamin D, calcium, parathyroid hormone, hormone replacement therapy, bisphosphonates, risedronate, alendronate, etidronate, raloxifene, zoledronic acid, calcitonin, and teriparatide. A manual search of the bibliographies of retrieved articles and information from clinicians in the field was used to supplement this. The search was limited to studies published in English-language journals.

Studies were retained for review when participants were older adults living in LTC or nursing home settings and when fracture rate was examined. Six studies met these criteria. Two of these studies3,4 were conducted with the same study sample, and both were used to support the 2002 and 2006 clinical guidelines. For each of these studies, information was gathered on the number, sex, and mean age and SD of participants in the intervention groups; inclusion and exclusion criteria; and mobility status.

RESULTS

Review of Studies Used to Support the 2002 and 2006 Clinical Guidelines

Of the 50 studies reviewed, 48 (96.0%) were conducted with community-dwelling participants. Some studies (n=14; 29.2%) specified that they included only healthy, ambulatory participants.5–18 Others excluded those with dementia,19 functional impairment,20 poor or fair health,21 or evidence of present or past disease, such as cancer or psychiatric or organic disease.13,17,22–26 Two studies3,4 were conducted with women living in nursing homes, but these were conducted with the same study population, the latter study being an extension of the former. The majority of studies were conducted solely with women (n=44; 88.0%), three (6.0%) with men,26–28 and three (6.0%) with men and women.16,29,30

Table 1 presents the composite mean age for the trials on bisphosphonates, calcitonin, calcium and vitamin D, HRT, and PTH that were included in this review. Also presented is the number of studies in which 1 SD above the mean fell above the age of 85. Assuming a normal distribution, 84.0% of the study participants will be below the mean plus 1 SD. The mean age of women in institutional care is approximately 85±7 and of men is 82±8.57 Consistent with this, the mean age of 283 institutionalized women experiencing hip fracture in a Canadian study was 86.3±6.2.58 The mean age of women living in LTC (85) has been used as the index age in Table 1, because the majority of participants in the studies reviewed were women. All of the studies conducted on calcitonin, HRT, and PTH were conducted with unique patient populations. Two of the studies conducted on calcium and vitamin D employed the same study sample, the second study being an extension of the first.3,4 Eighteen (58.1%) of the 31 studies conducted on bisphosphonates were conducted with original patient populations.10,14–18,23,24,26,30,39,42–47 The remaining bis-phosphonate trials were conducted with the same sample, a subsample of an originally reported study sample, or populations pooled from more than one similar study.

Table 1. Composite Mean Age Distribution for the Various Medication Trials and Trials Conducted with Institutionalized Participants
Trial n Composite Mean
Age
n
Studies
Reviewed
Studies that
Provided Mean
Age*
Studies that
Provided
SD for Mean Age
Studies in Which the Mean
Age Plus
1 SD Was ≥85
Bisphosphonates10–18,23–26,31–47 31 30 69.9 26 0
Calcium and vitamin D3,4,6,7,20,21,29,48 8 8 79.0 8 2
Calcitonin8,9,22,49 4 2 56.0 2 0
Parathyroid hormone5,27,28 3 3 66.6 2 0
Hormone replacement therapy19,50,51,52 4 2 62.9 1 0
Trials conducted with institutionalized participants3,4,53–56 6 5 84.1 5 5
  • * This represents the number of studies that were included in the calculation of the composite mean age.
  • The mean age of women in living in long-term care (85),57 because the majority of participants in the studies reviewed were women.
  • SD=standard deviation.

Not all of the studies provided the mean age and SD of participants. Four studies (8.0%) did not provide the mean age of participants.9,11,22,52 One study provided the median age.51 Eleven studies (22.0%) did not provide SDs.9,11,12,22,26,32,34,41,50–52

Across all of the trials, mean ages ranged from 51.5 (SD not available)50 to 84±6;3,4 the two studies with the highest mean age were conducted in nursing homes. When these studies are removed from the analysis, the highest mean age is 77.3±6.20 For the majority of studies (33/48; 68.8%) the mean age was younger than 70. Twelve studies3,4,6,10,20,21,23,29,30,37,47,48 had mean ages older than 70, two of which were conducted with the same sample.3,4

For bisphosphonate trials, the composite mean age was 70 (Table 1); the mean age of participants in the individual studies ranged from 52.7±1.018 to 76.4±8.6.23 For the majority of these studies (24; 77.4%), the mean age was younger than 70.12–18,24–26,32–36,38–46 In none of the bisphosphonate trials did 1 SD above the mean fall above age 85.

For the calcium and vitamin D trials, the composite mean age was 79 (Table 1); the mean age of participants in the individual studies ranged from 58±4.07 to 84±6.0.3,4 For the majority of these studies (7; 87.5%), the mean age was older than 70.3,4,6,20,21,29,48 One SD above the mean was older than 85 in two of the eight (25.0%) calcium and vitamin D trials.

For the trials of the remaining agents,5,8,9,19,22,27,28,49–52 the composite mean ages ranged from 56 to 67 (Table 1); for all of these trials, the mean ages of participants were younger than 70 (SD=3.98 to 8.127). For the calcitonin trials, only two studies8,49 provided information on the mean age of participants (55.6±3.9 and 68.4±7.6, respectively). For the HRT trials, the mean age of participants in the individual studies ranged from 51.5 (no SD provided)8 to 63.2±7.1.19 For the PTH trials, the mean age of participants in the individual studies ranged from 54.5±8.127 to 69.7±7.1.5 In none of these trials was 1 SD above the mean older than 85.

Review of Studies Conducted with Institutionalized Participants

Each of the six studies conducted with participants living in LTC examined the efficacy of calcium and vitamin D.4,53–56 One study was a controlled trial of a multimodal quality improvement intervention that included an analysis of whatever agent residents were prescribed, namely, bisphosphonates, calcium and vitamin D, calcitonin, and HRT.54 All of these studies were conducted with original or unique resident populations.

Four of the five studies conducted with institutionalized participants included only residents who were ambulatory or mobile, able to transfer independently, or not permanently bedridden.4,53,54,56 Only one study included residents with mobility, cognitive, visual, hearing, or communication impairments.55

The composite mean age for the trials conducted on institutionalized participants is presented in Table 1. The composite mean age was 84; the mean age of participants for the individual studies ranged from 83 (no SD provided)54 to 85.2±7.1.53 In all of the five trials conducted with institutionalized participants that provided a mean age, 1 SD above the mean was older than 85.

DISCUSSION

This analysis clearly shows that the frailer segment of the population is neglected in clinical trials of osteoporosis fracture prevention even though the most significant fracture of osteoporosis, hip fracture, is predominantly one of older adults and in particular frail older adults, many of whom live in LTC settings. There is a marked contrast in age between participants in the studies conducted in institutions and the age of those in the studies conducted in the community.

Most evidence collected through randomized trials is derived from a selected group of the population of interest. In recent years, the problem of clinical trial participants not being representative of clinical practice patients, and in particular older patients being relatively excluded, has become recognized.59 In the case of frail older people, all would have been excluded from the vast majority of the osteoporosis trials. In only one case was the study in the community clearly directed to the older end of the spectrum, and that study failed to show significant benefit from the treatment in the prevention of hip fractures.23 No antiresorptive drug fracture prevention trials have been performed in LTC. In the various fracture outcome studies performed in the community, the inclusion and exclusion criteria, when they were stated, would have excluded all LTC residents, as well as frailer older adults living in the community. Only the studies of vitamin D come close to addressing the prevention of fractures in this population, and even in those, the inclusion of only mobile participants excludes a significant proportion of those living in LTC.53

There are no clear guidelines that govern the generalizability of research findings. For generalizability to be justified, one would expect the study group to be representative of the population at large in major ways. When the population of interest is LTC residents, the study population should show a similar level of frailty, falling risk, dementia, and limited mobility, all of which may affect the rate and nature of falling. Extrapolation of study findings by extending the evidence to individuals who are members of a different population or at least a subpopulation that differs significantly from the study population may not be justifiable. For example, positive findings in women do not necessarily mean the same outcome for men. Nor do findings in the young, middle aged, and young older populations necessarily predict what will happen in the frail older population. The latter differ from the former in many ways that could be considered relevant (e.g., frailty, falls, nature of falls, padding).

It has been noted that it is the demonstration of effectiveness across a range of settings and populations that constitutes generalizability.60 Another study provided a list of things that can affect external validity (e.g., the setting of the trial, participant selection, characteristics of randomized participants, differences between the trial protocol and routine practice, outcome measurement and follow-up, and adverse effects) and that should be reported in published trials to help judge their generalizability to certain settings and patient groups,61 but details are rarely reported. In the present context, reasons for excluding frail older adults from clinical trials could include, for example, cognitive impairment or the inability to get to the study center because of limited mobility, features that would apply to many, if not most, residents in LTC. Generalizability is then suspect, because frail older adults may not show the same response to treatment when managed clinically like younger trial participants.

Selection bias, as reflected in selection criteria, and enrollment bias are significant methodological concerns. For example, although the risedronate trials enrolled people up to the age of 85, and fracture prevalence rises with age, the average age of the final group was just 71±7 in one study and 69±7.1 in the other.10,35 Studies that exclude patients on the basis of their mental or physical functional shortcomings or disease load will exclude many frail, older individuals. Requirements to be mobile and to attend clinics for assessment and test, will exclude many people aged 80 and older and persons who are housebound or in institutional settings.

The diagnosis of osteoporosis requires comment. This has evolved over the years from a clinical syndrome of fractures to one that includes bone density to, more recently, a risk-based model. It has become recognized that bone density does not fully explain the concept of osteoporosis and fracture risk, which may include bone quality. Nonetheless, most of the studies recruited participants on the basis of bone density with or without spinal fractures. Whether the concept of bone quality affects only susceptibility to spinal fractures or may also have an effect on hip fractures is not clear, but older subjects without low bone density, who may or may not have poor bone quality, seem to benefit much less from the standard osteoporosis treatment regimens in terms of reduction in hip fractures.

There is evidence that hip and other nonvertebral fractures can occur in the absence of thin bones as measured by bone mineral density (BMD). One study,62 using peripheral bone density measurements, found that, of those who sustained an osteoporotic fracture within a year of bone density measurement, 82% had a BMD above the t-score of −2.5 threshold, whereas another63 found that 54% of those who fractured a hip and 74% of those with any nonvertebral fracture, did not have low BMD (below −2.5 t-score) as measured by central BMD at the hip site. (T-score is the number of SDs an individual's bone density is below or above the average bone density of young people at maximum bone mass. A t-score of −2.5 has previously been used to define osteoporosis.) If this is the case, then treatment with a medication that has been shown to reduce fractures in those with fragile bones, as defined by low bone density, might not be expected to have much effect in reducing fractures in those who do not have low bone density. This may be why selection for antiresorptive treatment on the basis of criteria other than low BMD fails to reduce nonvertebral fracture risk, as shown by the risedronate study of women aged 80 and older selected for risk factors other than low BMD.23 When the oldest women without osteoporosis were examined, the hip fracture rate was 5.6% in the 3 years, compared with 9.7% in those with a t-score of −2.5 or worse. In the latter case, risedronate reduced the rate to 7.2%. Looked at simplistically, 4.1% of the risk of hip fracture might be considered to be due to osteoporosis, and this is reduced by an absolute 2.5%, that is, 40.0% by treatment. This is the same reduction as seen in the younger group with osteoporosis.

There are two major contributing factors for fractures (osteoporosis and falling), and the relevant contribution of these may vary according to age, setting, frailty, and comorbidities, especially cognitive status. An osteoporosis treatment's effectiveness may well be reduced as falling, and the changing nature of the fall, become more important determinants of the fracture.

There is evidence that there are important age-related differences in how people fall.64 Young people fall about half as often as old people, and they fall in a different way; falls in young people are more likely to be in a faster forward motion, and the sideways fall that characterizes falls in older adults is uncommon.65,66 Although the skeleton is designed to withstand a fall from a standing height in a young person, the skeleton of an older person may not be able to withstand the fall to which old people are susceptible. The fall onto the outstretched hand, particularly backward, appears to generate enough force to break most wrists.67 A lateral fall onto the greater trochanter, the fall that characterizes older adults and is the cause of the hip fracture, has been calculated to generate enough force to break most hips, and it is this type of fall that seems to be particularly difficult to protect oneself during, regardless of age.65,66,68 These “unprotected falls” are much more traumatic to the skeleton of older persons than the falls of young people. With respect to the risk of hip fracture, a fall that directly impacts the greater trochanter may break most hip bones.69 The more fragile the bone, the more likely it is that a fall that does not directly impact the greater trochanter will be sufficient to exceed the fracture threshold and still lead to a fracture, so bone loss will increase the risk of fracture. The chances of a fall directly impacting the greater trochanter will probably increase with greater frequency of falling. It may be relevant, therefore, to note that falls in LTC affect from 50% to 75% of residents per year, well above the community rate of 25% to 30%.70 Consistent with this, one study58 found that, when corrected for sex and age distribution, the hip fracture rate in LTC was approximately 1.8 times as high as expected.

The incidence of dementia may also contribute to our understanding of falls and fracture rates, because it is an established risk factor for falling, and hip fractures have been found to be 6 times as high in older adults with cognitive impairment.71–73 Limits on multitasking ability, (the ability to pay attention to more than one thing at a time, a skill that may prove to be important not only in preventing falls, but also in allowing fallers to react and protect themselves) partly explain the way older people fall; this ability is compromised in people who are cognitively impaired.74,75 The prevalence of dementia in LTC homes is high, rises with age, and is more common in women, features that parallel the fracture data.

Overall, the prevention of spontaneous vertebral fractures is more likely to be achieved than of partly traumatic nonvertebral fractures. Vertebral fracture usually occurs when an individual is performing normal activities that the spine may be expected to withstand, such as bending and lifting, and when the strength of the vertebra has just dropped below the fracture threshold for the activity performed. Improving the strength of the vertebra may help keep the strength above the fracture threshold. In contrast, an older person's fall on the greater trochanter will, it appears, exceed the fracture threshold several-fold, and a marginal increase in strength is unlikely to prevent a fracture other than one caused by an indirect blow. As noted earlier, fracture can occur despite good BMD.62,63 If osteoporosis is a minor contributor to the risk of hip fracture, then the lower effect of its prevention than on the prevention of spinal fracture is understandable. When identified on the basis of frailty and risk of falling, the effect of bisphosphonates was insignificant as in a risedronate study.23 Likewise when recruited on the basis of the hip fracture itself, and not bone density criteria, and when most of the patients had a BMD above a T-score of −2.5, the subsequent effect of the bisphosphonate, this time zoledronic acid, on subsequent vertebral fracture was significant, whereas the effect on hip fracture rate was insignificant, although the study was not powered for the latter outcome.30 Another study76 found that treating patients with hip fracture with bisphosphonates reduces the incidence of a second hip fracture but only in those younger than 80, with those aged 80 and older showing no benefit. This finding is consistent with a changing etiology after this age. It has been suggested that a decline in bone quality may be a concern with aging and the increased susceptibility to fracture77 and so account for the greater incidence of hip fracture in these people. Alternatively, bisphosphonates, which are reputed to improve bone quality, appear still to be of limited value in preventing hip fractures in very old people.78

There are data to confirm that, in the institutionalized population, bone density predicts fractures.79,80 No doubt osteoporosis and the high fall rate both contribute to the high rate of hip fracture in this population. It may be that falling is the major determinant of the prevalence of hip fracture, whereas bone strength, however defined, affects individual risk. Although it remains to be demonstrated, preventing loss of bone strength with aging may reduce hip fracture risk somewhat, but it is likely that preventing falling, if it were possible, would essentially eliminate hip fracture altogether.

In conclusion, there is little evidence that the use of osteoporosis medication, beyond vitamin D, which may reduce falling, will reduce the hip fracture risk in frail older people and in particular frail older people in LTC. It appears that falling becomes a more important factor in the fracture causation than low bone density in an older and older population. Current evidence is insufficient to allow the application of the community-based guidelines to the institutionalized population, and studies directly involving this population, however difficult, are needed.

ACKNOWLEDGMENTS

Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper.

Author Contributions: Richard G. Crilly: study concept and design, review and analysis, interpretation, manuscript drafting, critical manuscript review, final manuscript approval. Loretta M. Hillier and Michelle Mason: completion of literature searches, quantitative data extraction, and analysis, manuscript drafting, critical manuscript review, final manuscript approval. Iris Gutmanis: study concept and design, interpretation, critical manuscript review, final manuscript approval. Libby Cox: study concept and design, critical manuscript review, final manuscript approval.

Sponsor's Role: There was no sponsor for this project.