We previously demonstrated that a case manager intervention improved osteoporosis (OP) treatment within 6 months of hip fracture compared with usual care. The second phase of the randomized trial compared a less intensive intervention, facilitated bone mineral density (BMD) testing, with usual care and the case manager intervention.
We initially randomized 220 hip fracture patients to either an OP case manager intervention or usual care. After completing the original trial at 6 months postfracture, usual care patients were reallocated to facilitated BMD testing; BMD tests were arranged and results sent to primary care physicians. Main outcomes (bisphosphonate treatment, BMD tests, receipt of appropriate care) were reascertained 1 year following hip fracture and compared with outcomes achieved by the OP case manager intervention and usual care.
Compared with usual care, facilitated BMD testing increased testing from 29% to 68% (P < 0.001), bisphosphonate use from 22% to 38% (P < 0.001), and receipt of appropriate care from 26% to 45% (P < 0.001). The more intensive (70 versus 30 minutes) and expensive ($56 versus $24 Canadian per patient) OP case manager intervention led to significantly higher bisphosphonate use (54% versus 38%; P = 0.03), receipt of appropriate care (71% versus 45%; P < 0.001), and more BMD testing (80% versus 68%; P = 0.06) than usual care followed by facilitated BMD testing.
Compared with usual care, 2 different inexpensive interventions resulted in significant increases in appropriate management of OP after hip fracture. The magnitude of improvements achieved was directly related to the intensity of the interventions.
Hip fracture is the most significant complication of osteoporosis (OP), with substantial morbidity, mortality, and risk of recurrent fractures (1–6). Unfortunately, diagnosis and treatment of OP in this population remains less than 10–20% (1–9). Although diverse interventions after fracture have had various degrees of effectiveness (3), to our knowledge few studies have examined hip fracture (10–13). Further, in most cases the true effectiveness and costs of the interventions compared with usual care have not been determined in randomized controlled trials, leading to justifiable hesitation of widespread adoption of treatment practices by health providers.
We therefore designed a randomized controlled trial of an OP case manager intervention versus usual care in patients with hip fracture (14). The case manager was empowered to contact the patient at 3 months following hip fracture, schedule a bone mineral density (BMD) test, have a study physician prescribe bisphosphonate therapy when appropriate, and have the prescription sent to the patient's pharmacist. We found that 6 months after fracture, the case manager intervention resulted in 51% of patients receiving bisphosphonate therapy (versus 22% of usual care controls) and 67% of patients receiving appropriate care (versus 26% of controls). At the end of the initial followup period, we tested a less intense but perhaps more widely applicable intervention for those patients who had been initially allocated to receiving usual care in the first 6 months following hip fracture: specifically, BMD tests were arranged and the results sent to the patients' primary care physicians (facilitated BMD testing).
The primary purpose of the second phase of the study was to determine the effectiveness of the facilitated BMD testing intervention compared with 1) usual care and 2) the OP case manager intervention.
SUBJECTS AND METHODS
The trial received Health Research Ethics Board approval. The detailed design and description of the study population and 6-month postfracture results have previously been reported (14). To meet enrollment criteria, subjects had to be age ≥50 years, able to give (or have a proxy provide) informed consent, and have no contraindications for bisphosphonate therapy. Patients who were unable to consent, already receiving bisphosphonates, had a pathologic fracture, or were residing in long-term care facilities were excluded.
In summary, 220 patients were randomized to receive either usual care (educational material on OP, fall prevention, and calcium and vitamin D supplementation) or to an intervention consisting of a case manager who arranged for BMD testing (by dual x-ray absorptiometry [DXA]) followed by further counseling and prescription of a bisphosphonate (alendronate 70 mg or risedronate 35 mg weekly) if the patient had low bone mass (less than or equal to −1.5 SD T score at the spine, total hip, or femoral neck sites). The primary outcome at 6 months following hip fracture was the number of patients receiving bisphosphonate therapy. Secondary outcomes included a composite outcome that designated guideline-concordant appropriate care, as well as rates of BMD testing. Appropriate care was defined as a BMD test performed and bisphosphonate treatment provided to those with low bone mass according to published guidelines current in 2004 when the study commenced (15, 16). Bisphosphonates were not prescribed if BMD was normal.
At 6 months after the hip fracture, we tested a second intervention that we termed facilitated BMD testing; this secondary intervention was planned a priori when we designed the initial phase of the randomized controlled trial using a pre–post study with concurrent control group study design. The facilitated BMD testing intervention was far less labor intensive and consisted of study personnel contacting all usual care patients at 6 months following fracture. If BMD testing or prescription of OP therapy had not yet commenced, the case manager arranged for BMD testing, with results being sent to the primary care physician for further management. At 12 months after hip fracture, all patients in both groups (usual care/facilitated BMD testing and case manager intervention) were contacted to determine the same outcomes as in the primary study (Figure 1).
For patients lost to followup due to withdrawal or death (n = 35), last known values were carried forward (Figure 2). We determined the primary outcome (proportion treated with bisphosphonates) in 2 comparisons: a within-group comparison of facilitated BMD testing with usual care using McNemar's test, and a between-group comparison of usual care/facilitated BMD testing with the original OP case manager intervention using chi-square tests.
The within-group comparison contrasted the proportion of usual care patients who received bisphosphonate treatment in the initial 6 months with the proportion of usual care patients who received bisphosphonate treatment in the second 6 months after receiving the facilitated BMD testing intervention.
Because only 3 patients in the case manager intervention group commenced treatment between 6 and 12 months following hip fracture (P = 0.25), suggesting that minimal treatment occurred in the second 6 months following hip fracture, the case manager intervention group and the usual care/facilitated BMD testing group were compared at 12 months to determine the impact of combined usual care and facilitated BMD testing intervention versus the original case manager intervention (Figure 1).
We also compared rates of appropriate care and BMD testing using these same methods. Readmissions, refractures, and death rates were also examined at 12 months. To estimate the costs of the facilitated BMD intervention, we undertook detailed time-motion studies in a random sample of 15 patients, analogous to those performed in the case manager intervention (14).
Study participation and patient flow are shown in Figure 2. The baseline characteristics of the 220 subjects have been previously reported (14), and an abbreviated description is listed in Table 1. In summary, the median age of the 220 patients was 75.9 years (interquartile range 62.9, 84.4 years), 142 (65%) were women, and 77 (35%) had ≥4 comorbidities.
Table 1. Baseline characteristics of 220 intervention and control patients with a hip fracture*
Intervention (n = 110)
Control (n = 110)
Values are the number (percentage) unless otherwise indicated.
Age, mean ± SD years
72.9 ± 11.7
73.4 ± 11.1
Weight <57 kg
Osteoporosis risk factors
Previous history of osteoporosis
Previous bone mineral density examination
Taking calcium and/or vitamin D
The facilitated BMD testing intervention led to a significant increase in the proportion of patients receiving bisphosphonate therapy compared with usual care at 6 months following hip fracture (Figure 3). Bisphosphonate therapy was prescribed for 24 (22%) patients receiving usual care during the first 6 months; this increased to 42 (38%) patients receiving therapy following facilitated BMD testing, an absolute increase of 16% (P < 0.001). Fifty-nine (54%) subjects who underwent the OP case manager intervention received bisphosphonate therapy compared with 42 (38%) subjects who underwent usual care followed by facilitated BMD testing (P = 0.03) (Table 2, Figure 3).
Table 2. Rates of treatment and testing for OP in 220 patients 12 months after a hip fracture, comparing 2 interventions for improving OP diagnosis and treatment*
BMD testing significantly increased following the facilitated BMD testing intervention compared with usual care (Figure 3). After facilitated BMD testing, a total of 75 (68%) patients underwent BMD testing compared with 32 (29%) usual care patients, an absolute increase of 39% (P < 0.001). In the case manager intervention group, 88 (80%) patients underwent BMD testing compared with 75 (68%) patients who underwent the less intensive intervention (P = 0.06) (Table 2, Figure 3).
Following facilitated BMD testing, appropriate care significantly improved compared with usual care (Figure 3). Forty-nine (45%) patients received appropriate care using the facilitated BMD approach compared with 29 (26%) patients following usual care, an absolute increase of 19% (P < 0.001). The case manager intervention led to 78 (71%) patients receiving appropriate care compared with 49 (45%) patients who received the facilitated BMD testing intervention (P < 0.001) (Table 2, Figure 3).
New fractures (n = 10) and hospital readmissions (n = 42) over the initial 12 months after fracture were not significantly different (P > 0.60) between the usual care/facilitated BMD testing group and the case manager intervention group. The facilitated BMD testing intervention took a median of 30 minutes to perform at a cost of $24 Canadian per patient compared with a median of 70 minutes at a cost of $56 Canadian per patient for the case manager intervention.
By taking advantage of a previously reported randomized trial and reallocating our original control group to facilitated BMD testing alone, we were able to compare the relative efficacy and direct costs of 2 interventions of differing intensity for the treatment of OP following hip fracture. We found that simply providing BMD results to primary care physicians resulted in a significant increase in the prescription of bisphosphonate therapy and delivery of appropriate care compared with usual care following a hip fracture (Figure 3). However, facilitated BMD testing was still considerably less effective than an intervention in which BMD was not only arranged, but a case manager was also able to arrange prescriptions for bisphosphonate therapy according to established OP guidelines (Table 2, Figure 3).
In terms of appropriate care as we defined it, we found that a case manager led to a treatment rate of 71%, compared with a usual care treatment rate of 26%. The facilitated BMD intervention augmented usual care by an additional 19% in the second 6 months following hip fracture (Figure 3). We have no reason to believe that this incremental benefit could not have been achieved if the BMD tests had been facilitated during the original hospitalization. Indeed, an analysis of the changes in the OP case manager arm of our trial in the second 6 months of the study suggests minimal OP-related management activity in the community 6 months or later post–hip fracture. In our study, more men were randomized to the intervention group. Other studies have shown that men receive lower rates of treatment following fracture (17). Thus, the allocation of patients in our study led to most patients in the undertreated group receiving appropriate OP management far surpassing usual care for men following hip fracture.
In the context of our original randomized trial, where a case manager was able to provide prescriptions to those patients with hip fracture who had low BMD, it is not readily apparent why primary care physicians had a lower rate of treatment after receiving the BMD results. Other studies have reported barriers to OP treatment that include (but are not limited to): doctors being too busy in the hospital, reluctance of physicians to take on additional responsibilities, patients being too ill or unwilling or unable to do DXA, patient preferences in therapy and knowledge of OP, lack of access to or reimbursement for DXA, conflicting guidelines, and fear of polypharmacy and adverse drug side effects (12, 18–22). These reasons, however, do not appear to be likely explanations in our study. Both interventions took place following hospital discharge, eligible patients were relatively healthy in the context of the general hip fracture population, and the case manager gave the patient education regarding the importance of diagnosing and treating OP. Cost is generally not a barrier to prescription in Alberta, Canada, because alendronate and risedronate (the 2 antiresorptives prescribed) are fully covered under the universal provincial drug plan available for this patient group.
Even our more intensive intervention resulted in one-third of patients still not receiving appropriate care within 1 year of hip fracture. One other valid randomized controlled trial of hip fracture patients that provided educational materials and reminders found an increase in BMD testing and/or bisphosphonate prescription rates from 7 (19%) of 40 patients to 15 (42%) (10). More work is required to determine how OP treatment can be maximized following a hip fracture.
Further, our patient group represents a community-dwelling, relatively healthy minority of the hip fracture group. The main reason for exclusion in the current trial was residence in a long-term care facility, and our results cannot be easily applied to that specific patient group. Alternate care models need to be developed to properly address this larger group of patients. One possible model is to simply treat all patients regardless of BMD, as was done in a recent study of a placebo-controlled trial of yearly zoledronic acid, which included any ambulatory patients who had an expected life expectancy of >6 months following their hip fracture (10). In that study, ∼10% of patients had normal BMD compared with 18.9% in our study, a difference that may reflect the age and increased health and mobility of our patient population. Subgroup analysis is not available from these authors to determine whether patients with normal BMD also had reduced fracture rates, which could justify treatment with antiresorptive therapy (10). This study also excluded long-term care patients. The 2008 National Osteoporosis Foundation Guidelines also recommend treating all fractures, but this entails treating 10–20% of individuals with normal BMD. Although this is a simple, practical approach, we question the advisability of treating such a large population unnecessarily until the safety and cost-effectiveness of such an approach could be demonstrated.
No previous quality improvement studies of OP-related interventions following hip fracture have reported direct costs, even in large-scale population health programs (23, 24). In our initial report, we demonstrated that the intensive case manager intervention could be done for $56 Canadian per patient. Facilitated BMD testing was less effective than a case manager, and, not unexpectedly, the health care costs for this less intensive intervention were also similarly lower (approximately half the cost of case manager intervention). Both costs appear to be very reasonable expenditures to ensure increased appropriate treatment rates, and we have verified this conclusion by a formal cost-effectiveness analysis of our original study (25). Also, Sander et al recently found that a fracture coordinator significantly reduced the cost of fragility fractures (26).
Several limitations of this work need to be considered. First, our usual care group received far more attention than true usual care patients do in the community, in that they received information about OP treatment at the time of study enrollment and so may have been primed for the facilitated BMD testing intervention. In our previous report, we indicated that the treatment rates attained in the usual care group were higher than had been previously documented in practice audits (14).
Second, we studied processes of care, not clinical outcomes such as fracture or mortality. The recent study of Lyles et al (10) has shown effectiveness in reducing refracture rates and mortality by prescribing an intravenous bisphosphonate once per year. Although our study is not large enough or of sufficient duration to adequately address refracture end points, Lyles et al (10) would suggest that either of our intervention programs may subsequently lead to reduced refracture rates, and meta-analyses have confirmed that postfracture treatment with bisphosphonates reduces subsequent fractures (27–29).
Third, although all of our results were derived from a randomized trial, the comparisons of facilitated BMD testing with usual care were within-group (i.e., there was technically no external control group). Similarly, although our comparisons of facilitated BMD testing with the case manager intervention were between-group, they were not strictly concurrent. The study design for the secondary intervention (pre/post intervention with concurrent control group) is, however, recognized as a valid study design that has methodologic rigor (30). Although these results are not randomized, we controlled for many potential confounders of study results. We carefully monitored all potential participants so that we are able to describe all inclusions and exclusions, providing a true denominator for the study. We also assessed the groups for similarity prior to comparing outcomes and had excellent followup (90% of available subjects) out to 1 year after hip fracture. Thus, the main threat to validity for either of these comparisons might be that of unmeasured temporal trends. With the 6-month delay in implementing the facilitated BMD testing intervention, we may not have an accurate estimate of its impact considering that 26% of this group had already received treatment at the commencement of the secondary evaluation. However, the broader literature in this area describes universally low rates of OP treatment in hip fracture patients with little if any improvement over time (31). More importantly, in the 6 months after our original trial finished we demonstrated that there were no significant changes in treatment in the community in the case manager intervention group.
In conclusion, facilitated BMD testing almost doubles the rate of appropriate OP care compared with usual care following hip fracture. Although it is half as costly per patient as the OP case manager intervention, it is also half as effective; however, both interventions are much better than usual care as it is practiced in most jurisdictions (1–9). It is of importance that we have demonstrated that, for modest costs, OP care can be significantly improved. The degree of outcome success clearly is proportional to the amount of effort and expenditure that a health care provider is willing to devote.
Dr. Morrish had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.