Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Nov 1.
Published in final edited form as: J Bone Miner Res. 2022 Sep 27;37(11):2103–2111. doi: 10.1002/jbmr.4693

Osteoporosis Medications Prevent Subsequent Fracture in Frail Older Adults

Tanchanok Chattaris 1,2, Gahee Oh 2, Natalia A Gouskova 2, Dae Hyun Kim 2,3, Douglas P Kiel 2,3, Sarah D Berry 2,3
PMCID: PMC9712267  NIHMSID: NIHMS1834187  PMID: 36168189

Abstract

Frailty is common in older adults with fractures. Osteoporosis medications reduce subsequent fracture, but limited data exist on medication efficacy in frail individuals. Our objective was to determine whether medications reduce the risk of subsequent fracture in frail, older adults. A retrospective cohort of Medicare fee-for-service beneficiaries was conducted (2014–2016). We included adults aged ≥ 65 years who were hospitalized with fractures without osteoporosis treatment. Pre-fracture frailty was defined using claims-based frailty index (≥ 0.2 = frail). Exposure to any osteoporosis treatment (oral or intravenous bisphosphonates, denosumab, and teriparatide) was ascertained using Part B and D claims and categorized according to the cumulative duration of exposure: none, 1–90 days, and > 90 days. Subsequent fractures were ascertained from Part A or B claims. Cause-specific hazard models with time-varying exposure were fit to examine the association between treatment and fracture outcomes, controlling for relevant covariates. Among 29,904 patients hospitalized with fractures, 15,345 (51.3%) were frail, and 2,148 (7.2%) received osteoporosis treatment (median treatment duration 183.0 days). Patients who received treatment were younger (80.2 vs 82.2 years), female (86.5% vs 73.0%), less frail (0.20 vs 0.22) than patients without treatment. During follow-up, 5,079 (17.0%) patients experienced a subsequent fracture. Treatment with osteoporosis medications for > 90 days compared to no treatment reduced the risk of fracture (HR 0.82, 95% CI 0.68–1.00) overall. Results were similar in frail (HR 0.85; 95% CI 0.65–1.12) and non-frail (HR 0.80; 95% CI 0.61–1.04) patients, but not significant. In conclusion, osteoporosis treatment > 90 days was associated with similar trends in reduced risk of subsequent fracture in frail and non-frail persons. Treatment rates were very low, particularly among the frail. When weighing treatment options in frail older adults with hospitalized fractures, clinicians should be aware that drug therapy does not appear to lose its efficacy.

Keywords: Subsequent fracture, frailty, osteoporosis, older adults

Introduction

Osteoporosis and osteoporotic fractures remain significant public health challenges worldwide1. In western counties, the lifetime risk of any osteoporotic fracture ranges between 40–50% for women and 13–22% for men2. A recent study from the European Union3 estimated 2.7 million new fragility fractures occurred annually— equivalent to 7,332 fractures/day. Hip fractures are the most common site of fragility fractures, followed by clinical vertebral and distal forearm/proximal humerus fractures3. These fractures are associated with significant morbidities in older adults, including hospitalization, long-term care placement, impaired quality of life, disability, and even death4.

Frailty is a progressive age-related decline in physiological systems, leading to extreme vulnerability to stressors and increased risk of adverse health outcomes5. Among community-dwellers, 10.7% of adults aged ≥ 65 years6 and 25–50% of adults aged ≥85 years are frail7. This condition is common in older adults with fractures and could increase the risk of subsequent fracture.8, 9 A large Spanish population-based cohort study of adults≥ 75 years10 found that severe frailty was associated with an increased risk of hip, vertebral, and all fractures (hazard ratio [HR] with all fracture: 2.36, 95% CI 2.20–2.53). While the risk of subsequent fracture is greatest immediately following the index event, the median time to subsequent fracture is 3.7 ± 2.5 years11, suggesting that most adults have adequate time for interventions to prevent subsequent fracture.

Pharmacological interventions show promise in reducing subsequent fractures; however, trials of osteoporosis agents included small numbers of patients with advanced age and frailty12–15. In a large Swedish registry study, Nordstrom et al.16 found that bisphosphonate use was associated with a 26% decreased risk of subsequent hip fracture and a 38% reduced mortality rate during a 3-year follow-up, with similar efficacy in persons above the age of 80 years. Surprisingly, less than 10% of individuals in this registry were prescribed osteoporosis medications following the index hip fracture. The infrequent prescribing of osteoporosis medications could be influenced by limited data on efficacy in frail individuals17 and concerns about potential side effects of medications.16

This study aims to determine whether the osteoporosis medications reduce subsequent fractures similarly in older adults who are either frail or non-frail. We hypothesized that osteoporosis treatment would be less effective in frail individuals. As a secondary aim, we examined whether all-cause mortality and all-cause re-hospitalization differed according to osteoporosis treatment.

Method

Study design and participants

We conducted a retrospective cohort study of a 7.5% random sample of 2014–2016 Medicare fee-for-service (FFS) beneficiaries with at least 1 month of Parts A, B and D coverage (n= 4,653,045).

From this sample, we included beneficiaries aged 65 years or older who were hospitalized with a fracture at any sites except the head, hands, and feet from July 2014 to December 2015. Fractures were defined using International Classification of Diseases (ICD)-9 and -10 codes. (See the list of diagnosis codes in Supplemental Table 1). We excluded persons without continuous Medicare Parts A, B, and D enrollment, and without at least one Part D claim in the 6-months before the index hospitalized fracture. We additionally excluded beneficiaries with any claims for any hospitalized fractures, osteoporosis treatment (defined below) or hospice claims in the 6-months before the index hospitalized fracture. Figure 1 describes the selection of the study cohort. Participants were followed from the admission date of the index hospitalized fracture until death, Medicare FFS part A or B disenrollment, or study end date (December 31st, 2016). Data used in this manuscript was obtained retrospectively and does not require ethics committee approval.

Figure 1:

Figure 1:

Open in a new tab

Selection of Medicare fee-for-service beneficiaries with hospitalized fractures for a study of subsequent fracture

Frailty

There are two widely used methods to identify frailty in clinical practice: the frailty phenotype, and the deficit-accumulation frailty index (FI). Both methods have been applied and validated across multiple populations and settings. In our large cohort of Medicare beneficiaries, frailty was ascertained using a validated algorithm18 (statistical program codes are available at https://dataverse.harvard.edu/dataverse/cfi) that estimates a deficit-accumulation FI from 52 ICD-9 diagnostic variables, 25 Current Procedural Terminology (CPT)-4 variables, and 16 Healthcare Common Procedure Coding System (HCPCS) variables ascertained in the 6 months before the index hospitalized fracture. The claims-based FI is correlated with both FI and frailty phenotype, gait speed, and handgrip strength from clinical assessment, and is predictive of future death, nursing home admission and healthcare utilization19. The FI ranges between 0–1, with higher values indicating a greater degree of frailty. We defined frail as persons with a FI ≥ 0.220.

Osteoporotic treatment

Osteoporotic treatment included oral and intravenous bisphosphonates, denosumab and teriparatide. Treatment was ascertained using Part B and D claims in the period between the index hospitalized fracture and the end of follow-up (i.e., subsequent fracture, death, or end of study). Treatment was categorized according to the cumulative treatment exposure. For oral medications, we used the exact days of supply from claims data to infer treatment exposure. For non-oral medications, we inferred the duration of exposure for each medication using the recommended prescribed frequency for persons with osteoporosis (See more details in Supplemental Table 3). Cumulative treatment exposure was divided into three groups: no exposure, 1–90 days, and > 90 days. Categorization of the exposure to osteoporosis treatment varied with time (see examples in Figure 2). If persons had a gap or discontinued treatment, the cumulative treatment exposure (1–90 days or > 90 days) categorization remained unchanged.

Figure 2: Definition and assignment of cumulative osteoporosis treatment exposure.

Figure 2:

Open in a new tab

This figure illustrates the examples of treatment group assignments in this study. Person 1 started osteoporosis medications on Day 60 after admission, so this person was assigned to the no exposure group during Day 0–60 and exposure 1–90 days groups during Day 61–120. Person 2 started treatment on Day 30 after admission and received a longer duration of treatment, so Day 120–180 was assigned to the exposure > 90 days group. Person 3 had a treatment gap between Day 90 and 120, but this will not change the cumulative treatment exposure time and remained in the same group after restarting the treatment.

Other characteristics

The patients’ baseline characteristics were ascertained from the Medicare Database. Data on age, sex, race, and Medicaid eligibility status were collected at the time of the index hospitalized fracture. For others (e.g., claims-based comorbidity score21, number of medications, antipsychotic use, hypnotics and sedatives use, oral glucocorticoid use and number of any hospitalizations), we identified characteristics of patients in the 6-months before the index hospitalized fracture using Part A, B and D claims. We also identified relevant comorbidities before the index hospitalized fracture (such as hypertension, diabetes mellitus, ischemic heart disease, parkinsonism, dementia, depression, rheumatoid arthritis, osteoarthritis of knee, osteoporosis, cataract, glaucoma, and malignancies) using standardized definitions from the chronic conditions data warehouse.

Outcome measurement

The outcome of interest was subsequent fracture at any site except the head, hands, and feet. Subsequent fractures were ascertained using inpatient and outpatient claims (Part A, Part B Carrier, and Part B Outpatient; see Supplemental Table 1). Fracture definitions were based on a validated ICD-922 and -10 algorithm. The subsequent fracture must have occurred ≥100 days apart from the index hospitalized fracture if at the same site to avoid potential misclassification from claims that represent complications from the index hospitalized fracture.

As secondary outcomes, we considered mortality from the Medicare beneficiary summary file and all-cause re-hospitalizations from Part A claims. All outcomes were followed from the date of the index hospitalized fracture and censored on December 31st, 2016, or Medicare FFS part A or B disenrollment date, whichever came first.

Statistical analysis

We described baseline characteristics according to whether patients were ever exposed or never exposed to osteoporosis treatment after index hospitalized fracture. Comparisons of categorical and continuous covariates were performed by the chi-square test and Student t-test, respectively. We used the cause-specific hazards approach (i.e., competing risk analysis) with time-varying exposure to examine the association between osteoporosis treatment and subsequent fracture, using no osteoporosis treatment as the reference group. Immortal time bias was handled using the Mantel-Byar method23. For the model with subsequent fracture outcomes, we adjusted for factors that were a priori associated with subsequent fracture including age, sex, race, hypertension, diabetes, dementia, depression, osteoporosis, number of medications, use of antipsychotics, hypnotics and sedatives, and number of hospitalizations. As an alternative modeling strategy, we applied Fine and Gray models to examine the association between osteoporosis treatment and subsequent fracture accounting for the competing risk of death24. We also calculated the E-value, which is the amount of unmeasured confounding in addition to the characteristics we already controlled for that would be needed to make our results null25.

For the secondary outcomes mortality and re-hospitalization, we used cause-specific hazards models that included Medicaid status, comorbidity score21, ischemic heart disease, heart failure, arrhythmia, parkinsonism, rheumatoid arthritis, anemia, chronic kidney diseases, chronic liver diseases, solid and hematologic malignancies, and types of index fracture. The models were repeated separately in the frail and non-frail groups. For all models, a 2-sided p-value < 0.05 was considered statistically significant. All analyses were conducted using SAS 9.4 (TS1M6) software package.

Subgroup analyses

For the primary outcome of subsequent fracture, we conducted several subgroup analyses, including advanced age group (≥ 80 years), restricting the index hospitalized fracture to hip, limiting the type of osteoporosis treatment to oral bisphosphonates, and subsequent hospitalized fractures identified by only using Part A claims.

Results

Of 29,904 beneficiaries with index hospitalized fractures, the mean age was 82.1 (± 8.5) years and 22,131 (74.0%) were female. Just over half of patients (n=15,345, 51.3%) were frail according to our criteria. Only 2,148 participants (7.2%) received any osteoporosis medications during follow-up. Baseline characteristics of patients at index hospitalized fracture according to treatment status are shown in Table 1. Patients who received osteoporosis medications were younger (80.2 VS 82.2), more likely to be female (86.5% VS 73.0%), more robust (claim-based frailty index 0.20 VS 0.22) as compared with participants who were never treated with osteoporosis medications. Diagnosis of rheumatoid arthritis, osteoporosis, and history of having used oral glucocorticoid medications was also more common in participants who received osteoporosis treatment. The most frequent incident index hospitalized fracture type was hip fracture (49.7%), followed by thoracolumbar vertebral fracture (13.9%) and pelvic fracture (10.3%). The most common osteoporosis medication prescribed was alendronate (57.5%), followed by denosumab (30.6%) and teriparatide (6.6%); see Supplemental Table 3. Median treatment time was 183.0 (Interquartile range [IQR] 84.0–366.0) days.

Table 1.

Baseline characteristics of participants at index hospitalized fracture

Baseline characteristics Exposed to osteoporosis medications before subsequent fracture* (n=2,148) Not exposed to osteoporosis medications (n=27,756) p-value
Age, mean (SD) 80.2 (8.0) 82.2 (8.5) <0.01
Female, n (%) 1,859 (86.5) 20,272 (73.0) <0.01
Race, n (%)
 White 1,921 (89.4) 24,829 (89.5) 0.04
 Black (African-American) 59 (2.7) 1,035 (3.7)
 Hispanic 101 (4.7) 1,116 (4.0)
 Others 67 (3.1) 776 (2.8)
Claim-based frailty index, mean (SD) 0.20 (0.07) 0.22 (0.08) <0.01
 FI≥ 0.2, n (%) 929 (43.2) 14,416 (51.9) <0.01
Hypertension, n (%) 1,543 (71.8) 20,820 (75.0) <0.01
Diabetes, n (%) <0.01
 With insulin 153 (7.1) 2,365 (8.5)
 Without insulin 459 (21.4) 6,785 (24.4)
Dementia, n (%) 319 (14.9) 7,389 (26.6) <0.01
Depression, n (%) 562 (26.2) 7,695 (27.7) 0.12
Osteoporosis, n (%) 513 (23.9) 3,642 (13.1) <0.001
Number of medications, n (%) <0.001
 0–5 medications 401 (18.7) 5,685 (20.5)
 6–10 medications 600 (27.9) 8,132 (29.3)
 ≥11 medications 1,147 (53.4) 13,939 (50.2)
Antipsychotics use, n (%) 133 (6.2) 2,714 (9.8) <0.01
Hypnotics and sedatives use, n (%) 209 (9.7) 2,169 (7.8) <0.01
Types of index fracture, n (%)
 Hip fracture 1,068 (49.7) 13,795 (49.7) 0.99
Thoracolumbar vertebral fracture 349 (16.2) 3,196 (11.5) <0.01
 Pelvic fracture 246 (11.5) 2,536 (9.1) <0.01
 Upper arm fracture 137 (6.4) 2,150 (7.7) 0.02
 Other fractures† 494 (16.2) 7,520 (22) <0.01
Open in a new tab
*

Exposed and Not exposed categories represent convenient groupings to describe the cohort, whereas all models used a time-varying approach to categorize treatment (Figure 2). “Exposed” indicates a Part B or D claim for osteoporosis treatment between the index hospitalized fracture and the end of follow-up (i.e., subsequent fracture, death, or end of study). Note that participants with a claim for osteoporosis treatment in the 6-month preceding the index hospitalized fracture were excluded from the cohort (Figure 1).

†

Other fractures comprised of fractures at femoral shaft, patellar, tibia and fibula, ribs and sternal, scapula and clavicle, lower arm and wrist. More details are in supplementary table S2

Subsequent fracture

During a median follow-up of 469 (IQR 234–660) days, 2,731 (17.8% or 15.6 per 100 person-years [PY]) subsequent fractures occurred in frail and 2,348 (16.1% or 12.0 per 100 PY) occurred in non-frail participants. Figure 3 shows the effect estimates of treatment on subsequent fracture overall, and according to frailty subgroups.

Figure 3:

Figure 3:

Open in a new tab

Subsequent fracture outcomes overall and according to frailty status

Osteoporosis medication use for 1–90 days was associated with an increased rate of subsequent fracture in the overall and non-frail groups (adjusted hazard ratio [aHR] 1.24; 95% confidence interval [95% CI] 1.05–1.46 and aHR 1.31; 95% CI 1.04–1.65, respectively). Results were similar in frail patients (aHR 1.17;95% CI 0.93–1.49) but did not reach statistical significance. Receipt of osteoporosis medications for >90 days was associated with a reduced risk of subsequent fracture in all groups (overall: aHR 0.82; 95% CI 0.68–1.00; frail: aHR 0.85; 95% CI 0.65–1.12; and non-frail: aHR 0.80; 95% CI 0.61–1.04) but only reached statistical significance in the overall cohort (see Figure 3). Fine and Gray models produced similar results with effect estimates of treatment for 1–90 days generally higher and estimates for > 90 days closer to the null, as compared with the cause-specific hazard model results (Supplemental Table 4).

In the overall cohort, the E-values (hazard ratios) for the point estimate for subsequent fractures by osteoporosis treatment 1–90 days and > 90 days were 1.80 and 1.74, respectively.

Subgroup analyses showed similar trends in participants of older age, receiving oral bisphosphonates, with index hospitalized hip fracture and subsequent in-patient fractures only, although effect estimates were closer to the null in the frail group. For example, treatment with oral bisphosphonates for > 90 days was associated with a 17% (95% CI 0.58–1.19) decreased risk of subsequent fracture in non-frail patients compared with 13% (95% CI 0.61–1.26) in frail patients. Only treatment for > 90 days among non-frail patients with index hospitalized hip fracture reached statistical significance (aHR 0.61; 95% CI 0.40–0.92); see Figure 4.

Figure 4: Subgroup analyses: subsequent fracture outcomes according to frailty status.

Figure 4:

Open in a new tab

Dash line = Treatment exposure 1–90 days, Solid line = Treatment exposure > 90 days

Secondary Outcomes

During follow-up, 5,493 (35.8 % or 27.5 per 100 PY) of frail patients and 2,348 (16.1% or 11.4 per 100 PY) of the non-frail died from any causes. In fully adjusted models, treatment of ≤ 90 days was associated with a significant lower mortality rate in frail (aHR 0.61; 95% CI 0.48–0.77) but not for non-frail persons (aHR 0.89; 95% CI 0.68–1.15). Receipt of osteoporosis medications for more than 90 days was associated with decreased mortality in both groups (aHR 0.74; 95% CI 0.60–0.90 and aHR 0.68; 95% CI 0.52–0.89 for frail and non-frail, respectively); see Table 2.

Table 2.

All-cause mortality and any hospitalization during study period

Secondary outcomes 1–90 days treatment > 90 days treatment
Frail
HR (95% CI)		 Non frail
HR (95% CI)		 Frail
HR (95% CI)		 Non frail
HR (95% CI)
All-cause mortality 0.50
(0.40–0.64)		 0.75
(0.57–0.97)		 0.61
(0.50–0.75)		 0.58
(0.45–0.76)
All-cause mortality* 0.61
(0.48–0.77)		 0.89
(0.68–1.15)		 0.74
(0.60–0.91)		 0.68
(0.52–0.89)
Re-hospitalization 1.05
(0.91–1.22)		 0.82
(0.69–0.98)		 1.00
(0.85–1.17)		 0.95
(0.81–1.12)
Re-hospitalization* 1.06
(0.92–1.23)		 0.86
(0.72–1.03)		 1.00
(0.85–1.17)		 1.00
(0.85–1.17)
Open in a new tab
*

Adjusted with age, sex, race, comorbidity score, hypertension, diabetes, ischemic heart disease, heart failure, arrhythmia, dementia, parkinsonism, depression, anemia, chronic kidney diseases, chronic liver diseases, rheumatoid arthritis, solid and hematologic malignancies, number of medications, antipsychotics use, number of hospitalizations, Medicaid status and types of index fracture

When examining re-hospitalization risk during follow up, 9,310 (60.7 % or 78.8 per 100 PY) of frail and 6,505 (44.7 % or 41.0 per 100 PY) of non-frail persons were re-hospitalized. Osteoporosis treatment was not associated with hospitalization in either patients who were frail or not frail. (For treatment > 90 days, aHR 1.00; 95% CI 0.85–1.17 for both groups); see Table 2.

Discussion

In a large cohort of Medicare FFS beneficiaries with hospitalized fractures and not previously treated with any osteoporosis medications, we investigated the association of osteoporosis medications after index hospitalized fracture with subsequent fracture risk. Treatment rates with osteoporosis medications were very low, and in fact, half of the treated patients received no more than 6 months of treatment during follow-up. Nonetheless, we found that treatment with osteoporosis medications beyond 90 days reduced the risk of subsequent fracture with similar trends in both frail and non-frail persons. These findings support that drug therapy does not appear to lose its efficacy in older adults with hospitalized fractures.

Previous studies, including randomized controlled trials13, 26, 27, provide strong evidence that osteoporosis medications prevent subsequent fractures. For example, in the HORIZON study13, intravenous zoledronic acid reduced the risk of subsequent clinical fractures by 35%, and in a post-hoc analysis of FREEDOM and FREEDOM extension studies26, denosumab decreased risk of subsequent osteoporotic fractures in women with osteoporosis by 41% compared with placebo. Frailty was not measured directly in either trial, but frail persons may have been excluded. The HORIZON study required written informed consent from patients and proxies if cognitive impairment was detected, and FREEDOM did not give details about cognitive or functional impairment in their sample. Moreover, the mean age in the HORIZON study13 was 74 years (with just 13.8% aged ≥ 85 years) and for the FREEDOM trial, the mean age was 72 years, ten years younger than the mean age of 82.1 years in our study. Despite the strong evidence to support osteoporosis treatment for subsequent fracture prevention, many providers withhold treatment from frail older adults, often citing a lack of evidence that these drugs are effective. Given the increased risk of fractures associated with frailty, additional research is needed to examine the safety and efficacy of these medications in frail, older adults.

Similarly, observational studies have provided supporting evidence that osteoporosis medications reduce the risk of subsequent fracture. Nordstorm et al.16 showed the advantage of bisphosphonates in preventing subsequent hip fracture in a Swedish population aged > 80 years old with a mean treatment time of 1.6 years. Another large population cohort in Sweden28 found an increased rate of any new clinical fracture within 6-months of bisphosphonate initiation in older adults with clinical fracture, but the risk of subsequent fracture decreased with longer duration of bisphosphonate use (>12 months). With careful handling of immortal time bias, we similarly found an elevated risk of fracture when bisphosphonates were used for < 90 days. The elevated risk of subsequent fracture with short treatment duration is likely multifactorial. First, the risk of re-fracture is highest in the first year following a fracture29, 30, and those who are at the greatest imminent risk of fracture may be preferentially treated. Despite attempts to adjust for confounding in our models, residual confounding may remain. Moreover, the finding may be impacted by adherence. Many individuals stop osteoporosis treatment after a single prescription for oral medications31, and persons with poor adherence are at an increased risk of falls and fracture32, 33.

Previous studies of highly effective therapies found less benefit when treatment was offered to frail individuals. For example, Mallery et al.34 showed no benefit in statin therapy for primary prevention of cardiovascular outcomes in older adults who were severely frail. We are not aware of any other studies that have examined the effect of osteoporosis treatment by frailty status. Contrary to our hypothesis, our study suggests that osteoporosis medications may have similar efficacy in persons regardless of whether they are frail or non-frail.

We acknowledge that in our subgroup analyses the effect of osteoporosis treatment on subsequent fracture risk was less than previously mentioned studies13, 16, 28, particularly for patients with a prior hip fracture or patients treated with oral bisphosphonates. The median duration of treatment in our study was just 183 days, and it is possible that these medications are more effective in reducing subsequent fractures with longer duration of treatment. Additionally, nearly one-third of treated patients were prescribed denosumab in our study, and it is possible that denosumab differentially affects subsequent fracture and fall risk in frail individuals. In a post hoc analysis of the FREEDOM trial35, denosumab was associated with a prominent reduction of hip fracture risk in postmenopausal women age ≥ 75 years as compared with a more modest reduction in younger women. Additional research is needed to examine the safety and efficacy of denosumab in frail, older adults.

Our mortality analysis suggests that residual confounding may remain despite adjusting for many confounders, as we found that osteoporosis treatment for ≤ 90 days was associated with a reduced mortality, particularly in frail patients. Fine and Gray models are susceptible to bias in the presence of unmeasured mortality confounders, and so we believe our primary approach to modeling the effect of osteoporosis treatment on subsequent fractures to be more accurate36. We found that treatment for more than 90 days was associated with a modest survival benefit in both frail and non-frail older adults. These findings are consistent with previous studies13, 37–40 including randomized control trials that found bisphosphonates decrease mortality. There have been many hypotheses to explain this association beyond protection of subsequent fracture, including protective effects on the immune system and prevention of the release of cytokines and other toxins from harming bone41. Associations between nitrogen-containing bisphosphonates and reduced risk of myocardial infarction42 and pneumonia43 were also observed and could play a role in the survival benefits. Further study is needed to clarify the underlying mechanism.

There are limitations to our study. First, as mentioned above, because of our study design, there are considerable differences in baseline characteristics between treated and untreated persons and unmeasured confounding may remain. For example, we did not have information on calcium and vitamin D intake, known risk factors for fracture; however limited data to support their ability to reduce fracture incidence in clinical trials. While it is possible that calcium and vitamin D supplementation differed according to treatment duration, we suspect there may be other, unmeasured prognostic variables that are not readily assessed using claims data, e.g., mobility status after the index fracture. Nevertheless, based on the magnitude of the E-value, we concluded that it would have required a moderate amount of unmeasured confounding in addition to measured confounders for which we had already controlled for to make the results null. Thus, residual confounding is unlikely to explain our findings. Second, we ascertained medication use from Medicare claims and did not have a measure of adherence. The most prevalent medication in our study was alendronate, and adherence with this medication is typically low, potentially leading to some misclassification with longer durations of exposure. However, since we did observe a protective association of treatment, our results may actually have underestimated the true protective effect. Third, some misclassification is possible when using claims data to ascertain fracture. Misclassification would be expected to bias results to the null if non-differential in relation to the exposure status. We required a 100-day window before considering a subsequent fracture at the same site, to avoid counting re-hospitalizations which could artificially raise rates of subsequent fracture over short time frames29. Fourth, we did not have data on fractures that occurred before 2014. Fifth, the assigned treatment duration for some osteoporosis medications is conservative, particularly for zoledronic acid that may be dosed less frequently than 365 days. Treatment with zoledronic acid was rare (0.7%), and our categorization of exposure (i.e., treatment > 90 days) would consider beneficiaries with this drug as exposed regardless. Finally, this study primarily included white, Medicare beneficiaries, and results may not generalize to other groups.

There are several strengths to our study. Observational study designs offer the opportunity to examine the effect of medications in a real-world setting. We also used a well-structured, validated tool to measure frailty in our study that could be replicated in other studies of osteoporosis using claims data. Models accounted for the competing risk of death, and both cause-specific hazards models and Fine and Gray models, provided similar estimates. Lastly, we also dealt with immortal time bias that can occur in observational studies of treatment effectiveness when non-treatment periods are not counted with an effective statistical approach. Given that a future clinical trial of these osteoporosis treatments in frail older adults will not likely ever be conducted, our study approach adds to the evidence that these drugs are effective in this population.

Conclusion

In persons with a hospitalized fracture, osteoporosis treatment of more than 90 days is associated with similar trends in reduced risk of subsequent fracture in frail and non-frail patients. Given these results, clinicians may want to discuss efficacy with patients when considering osteoporosis treatment in frail older adults with hospitalized fractures and reasonable life expectancy.

Supplementary Material

supinfo

Acknowledgments

This work was supported by the National Institute of Health, National Institute on Aging (NIA), K24 AG070106, R01 AG045441 and R01 AG071809.

Disclosures

DPK acted as a data safety monitoring board member for Agnovos and a scientific advisory board member for Solarea Bio, Reneo, and Pzifer. DPK received grants from Amgen and Solarea Bio. DPK and SDB received royalties from Wolters Kluwer.

Footnotes

Our data use agreement with the Centers for Medicare & Medicaid Services prohibits sharing Medicare claims data with others.

These analyses were presented in part as an oral presentation at the American Geriatric Society meeting on May 12, 2022.

References

  • 1.Li G, Thabane L, Papaioannou A, Ioannidis G, Levine MAH, Adachi JD. An overview of osteoporosis and frailty in the elderly. BMC Musculoskelet Disord. 2017;18(1):46-. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Johnell O, Kanis J. Epidemiology of osteoporotic fractures. Osteoporos Int. 2005;16 Suppl 2:S3–7. [DOI] [PubMed] [Google Scholar]
  • 3.Borgström F, Karlsson L, Ortsäter G, Norton N, Halbout P, Cooper C, et al. Fragility fractures in Europe: burden, management and opportunities. Arch Osteoporos. 2020;15(1):59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Beaupre LA, Jones CA, Saunders LD, Johnston DW, Buckingham J, Majumdar SR. Best practices for elderly hip fracture patients. A systematic overview of the evidence. J Gen Intern Med. 2005;20(11):1019–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Cesari M, Prince M, Thiyagarajan JA, De Carvalho IA, Bernabei R, Chan P, et al. Frailty: An Emerging Public Health Priority. J Am Med Dir Assoc. 2016;17(3):188–92. [DOI] [PubMed] [Google Scholar]
  • 6.Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc. 2012;60(8):1487–92. [DOI] [PubMed] [Google Scholar]
  • 7.Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet. 2013;381(9868):752–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Rockwood K, Theou O, Mitnitski A. What are frailty instruments for? Age Ageing. 2015;44(4):545–7. [DOI] [PubMed] [Google Scholar]
  • 9.van den Bergh JP, van Geel TA, Geusens PP. Osteoporosis, frailty and fracture: implications for case finding and therapy. Nat Rev Rheumatol. 2012;8(3):163–72. [DOI] [PubMed] [Google Scholar]
  • 10.Middleton R, Poveda JL, Orfila Pernas F, Martinez Laguna D, Diez Perez A, Nogués X, et al. Mortality, Falls, and Fracture Risk Are Positively Associated With Frailty: A SIDIAP Cohort Study of 890 000 Patients. J Gerontol A Biol Sci Med Sci. 2022;77(1):148–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ruan WD, Wang P, Ma XL, Ge RP, Zhou XH. Analysis on the risk factors of second fracture in osteoporosis-related fractures. Chin J Traumatol. 2011;14(2):74–8. [PubMed] [Google Scholar]
  • 12.Wells GA, Cranney A, Peterson J, Boucher M, Shea B, Robinson V, et al. Alendronate for the primary and secondary prevention of osteoporotic fractures in postmenopausal women. Cochrane Database Syst Rev. 2008(1):Cd001155. [DOI] [PubMed] [Google Scholar]
  • 13.Lyles KW, Colón-Emeric CS, Magaziner JS, Adachi JD, Pieper CF, Mautalen C, et al. Zoledronic acid and clinical fractures and mortality after hip fracture. N Engl J Med. 2007;357(18):1799–809. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Wen F, Du H, Ding L, Hu J, Huang Z, Huang H, et al. Clinical efficacy and safety of drug interventions for primary and secondary prevention of osteoporotic fractures in postmenopausal women: Network meta-analysis followed by factor and cluster analysis. PLoS One. 2020;15(6):e0234123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Shi L, Min N, Wang F, Xue QY. Bisphosphonates for Secondary Prevention of Osteoporotic Fractures: A Bayesian Network Meta-Analysis of Randomized Controlled Trials. Biomed Res Int. 2019;2019:2594149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Nordström P, Toots A, Gustafson Y, Thorngren KG, Hommel A, Nordström A. Bisphosphonate Use After Hip Fracture in Older Adults: A Nationwide Retrospective Cohort Study. J Am Med Dir Assoc. 2017;18(6):515–21. [DOI] [PubMed] [Google Scholar]
  • 17.Zullo AR, Zhang T, Lee Y, McConeghy KW, Daiello LA, Kiel DP, et al. Effect of Bisphosphonates on Fracture Outcomes Among Frail Older Adults. J Am Geriatr Soc. 2019;67(4):768–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring Frailty in Medicare Data: Development and Validation of a Claims-Based Frailty Index. J Gerontol A Biol Sci Med Sci. 2018;73(7):980–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Kim DH, Patorno E, Pawar A, Lee H, Schneeweiss S, Glynn RJ. Measuring Frailty in Administrative Claims Data: Comparative Performance of Four Claims-Based Frailty Measures in the U.S. Medicare Data. J Gerontol A Biol Sci Med Sci. 2020;75(6):1120–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Strom JB, Xu J, Orkaby AR, Shen C, Charest BR, Kim DH, et al. Identification of Frailty Using a Claims-Based Frailty Index in the CoreValve Studies: Findings from the EXTEND-FRAILTY Study. J Am Heart Assoc. 2021;10(19):e022150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64(7):749–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Wright NC, Daigle SG, Melton ME, Delzell ES, Balasubramanian A, Curtis JR. The Design and Validation of a New Algorithm to Identify Incident Fractures in Administrative Claims Data. J Bone Miner Res. 2019;34(10):1798–807. [DOI] [PubMed] [Google Scholar]
  • 23.Mi X, Hammill BG, Curtis LH, Greiner MA, Setoguchi S. Impact of immortal person-time and time scale in comparative effectiveness research for medical devices: a case for implantable cardioverter-defibrillators. J Clin Epidemiol. 2013;66(8 Suppl):S138–44. [DOI] [PubMed] [Google Scholar]
  • 24.Austin PC, Lee DS, Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Circulation. 2016;133(6):601–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.VanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167(4):268–74. [DOI] [PubMed] [Google Scholar]
  • 26.Kendler DL, Chines A, Brandi ML, Papapoulos S, Lewiecki EM, Reginster JY, et al. The risk of subsequent osteoporotic fractures is decreased in subjects experiencing fracture while on denosumab: results from the FREEDOM and FREEDOM Extension studies. Osteoporos Int. 2019;30(1):71–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Jin YZ, Lee JH, Xu B, Cho M. Effect of medications on prevention of secondary osteoporotic vertebral compression fracture, non-vertebral fracture, and discontinuation due to adverse events: a meta-analysis of randomized controlled trials. BMC Musculoskelet Disord. 2019;20(1):399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Bergman J, Nordström A, Nordström P. Bisphosphonate use after clinical fracture and risk of new fracture. Osteoporos Int. 2018;29(4):937–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Johansson H, Siggeirsdóttir K, Harvey NC, Odén A, Gudnason V, McCloskey E, et al. Imminent risk of fracture after fracture. Osteoporos Int. 2017;28(3):775–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Balasubramanian A, Zhang J, Chen L, Wenkert D, Daigle SG, Grauer A, et al. Risk of subsequent fracture after prior fracture among older women. Osteoporos Int. 2019;30(1):79–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Curtis JR, Yun H, Matthews R, Saag KG, Delzell E. Adherence with intravenous zoledronate and intravenous ibandronate in the United States Medicare population. Arthritis Care Res (Hoboken). 2012;64(7):1054–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Berry SD, Quach L, Procter-Gray E, Kiel DP, Li W, Samelson EJ, et al. Poor adherence to medications may be associated with falls. J Gerontol A Biol Sci Med Sci. 2010;65(5):553–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Curtis JR, Delzell E, Chen L, Black D, Ensrud K, Judd S, et al. The relationship between bisphosphonate adherence and fracture: is it the behavior or the medication? Results from the placebo arm of the fracture intervention trial. J Bone Miner Res. 2011;26(4):683–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Mallery LH, Moorhouse P, McLean Veysey P, Allen M, Fleming I. Severely frail elderly patients do not need lipid-lowering drugs. Cleve Clin J Med. 2017;84(2):131–42. [DOI] [PubMed] [Google Scholar]
  • 35.Boonen S, Adachi JD, Man Z, Cummings SR, Lippuner K, Törring O, et al. Treatment with denosumab reduces the incidence of new vertebral and hip fractures in postmenopausal women at high risk. J Clin Endocrinol Metab. 2011;96(6):1727–36. [DOI] [PubMed] [Google Scholar]
  • 36.Barrowman MA, Peek N, Lambie M, Martin GP, Sperrin M. How unmeasured confounding in a competing risks setting can affect treatment effect estimates in observational studies. BMC Med Res Methodol. 2019;19(1):166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.van Geel T, Bliuc D, Geusens PPM, Center JR, Dinant GJ, Tran T, et al. Reduced mortality and subsequent fracture risk associated with oral bisphosphonate recommendation in a fracture liaison service setting: A prospective cohort study. PLoS One. 2018;13(6):e0198006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Sambrook PN, Cameron ID, Chen JS, March LM, Simpson JM, Cumming RG, et al. Oral bisphosphonates are associated with reduced mortality in frail older people: a prospective five-year study. Osteoporos Int. 2011;22(9):2551–6. [DOI] [PubMed] [Google Scholar]
  • 39.Abtahi S, Burden AM, Geusens P, van den Bergh JP, van Staa T, de Vries F. The Association of Oral Bisphosphonate Use With Mortality Risk Following a Major Osteoporotic Fracture in the United Kingdom: Population-Based Cohort Study. J Am Med Dir Assoc. 2020;21(6):811–6. [DOI] [PubMed] [Google Scholar]
  • 40.Center JR, Bliuc D, Nguyen ND, Nguyen TV, Eisman JA. Osteoporosis medication and reduced mortality risk in elderly women and men. J Clin Endocrinol Metab. 2011;96(4):1006–14. [DOI] [PubMed] [Google Scholar]
  • 41.Center JR. Fracture Burden: What Two and a Half Decades of Dubbo Osteoporosis Epidemiology Study Data Reveal About Clinical Outcomes of Osteoporosis. Curr Osteoporos Rep. 2017;15(2):88–95. [DOI] [PubMed] [Google Scholar]
  • 42.Sing CW, Wong AY, Kiel DP, Cheung EY, Lam JK, Cheung TT, et al. Association of Alendronate and Risk of Cardiovascular Events in Patients With Hip Fracture. J Bone Miner Res. 2018;33(8):1422–34. [DOI] [PubMed] [Google Scholar]
  • 43.Sing CW, Kiel DP, Hubbard RB, Lau WC, Li GH, Kung AW, et al. Nitrogen-Containing Bisphosphonates Are Associated With Reduced Risk of Pneumonia in Patients With Hip Fracture. J Bone Miner Res. 2020;35(9):1676–84. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supinfo
NIHMS1834187-supplement-supinfo.pdf (295.6KB, pdf)

RESOURCES