Exposure to Drugs Related to Injurious Falls Risk and to Bisphosphonates in Nursing Home Residents with Recent Fracture

Abstract

Objectives

To describe the frequency of use of drugs related to injurious falls risk (DRIFR) and oral bisphosphonates in nursing home (NH) beneficiaries with Medicare fee-for-service and Part D who had sustained a recent fracture.

Design

This is a cross-sectional study using data from 1/1/2017-6/30/2022. DRIFR included 18 medication classes. Use of DRIFRs and oral bisphosphonates were ascertained in the 30 days of NH care before a defined index date (day 31 for new residents; study start date for prevalent residents) using Part D claims.

Settings and Participants

We analyzed data from 14,559 NHs in the United States comprising 192,118 residents with a recent hip or other fracture, defined using a federally mandated clinical assessment (Minimum Data Set; MDS).

Methods

Patient characteristics were identified from the most recent MDS (≤92 days before index date). Descriptive statistics were calculated using t-tests, Wilcoxon rank-sum test, or Chi-Square tests. Patient characteristics and medication use were stratified by bisphosphonate prescriptions (yes/no).

Results

Of recent fractures, 41.4% of residents had hip fractures and 67.7% had fractures at other sites. Mean age was 83.8 years (range=65-110); 75.4% were female; 86.5% were non-Hispanic White. 18.8% of residents had a diagnosis of osteoporosis, which was more common in residents with bisphosphonate prescriptions than in residents without. Only 4.2% of residents had bisphosphonate prescriptions, while 72.8% had prescriptions of DRIFRs. Common DRIFRs included statins (31.1%), opioids (26.7%), newer anticonvulsants (22.1%), and loop diuretics (22.1%).

Conclusions and Implications

Use of DRIFRs is common whereas oral bisphosphonates are rarely prescribed in residents with a recent fracture receiving care in NHs. Our results suggest opioids and newer anticonvulsants may represent common candidates for medication review. Efforts to treat fractures and prevent secondary fractures should not only focus on increasing bisphosphonate treatment, but also on deprescribing DRIFR.

Brief summary:

Study of 192K nursing home residents with fractures: Only 4.2% received bone-protective bisphosphonates, while 72.8% received fall risk increasing drugs like benzodiazepines. Secondary fracture prevention efforts should include deprescribing.

Introduction

Each year, approximately 10% of community-dwelling adults experience a fall-related injury (FRI), and the incidence doubles for nursing home (NH) residents.^1,2 FRIs impact quality of life by way of functional decline and reduced mobility.³ Moreover, fractures increase subsequent fracture risk, so it is critical to consider interventions to reduce subsequent fracture.^4,5

NH residents are frequently exposed to medication classes that elevate FRIs risk, including psychotropics, certain antihypertensives, opioids, and others.^6-8 While these medication classes may be clinically indicated in NH residents with complex comorbidities, concurrent use of multiple FRI-associated medication classes may compound the likelihood of fracture. These medications may increase the risk of falls through sedation (e.g., benzodiazepines), impaired balance (e.g., barbiturates), orthostatic hypotension (e.g., tricyclic antidepressants), or other mechanisms, or they may increase injury risk or severity in the setting of a fall through effects on bone strength (e.g., glucocorticoids).^9-12 Additionally, osteoporosis treatments are associated with reduced incidence of fractures, yet these medications are infrequently prescribed in persons with fracture.¹³

Medication optimization in persons with fracture includes consideration of deprescribing drugs associated with FRI and prescribing osteoporosis treatment. Among residents receiving care in a NH following fracture, oral bisphosphonates are inexpensive and represent one component of guideline-recommended post-fracture medication optimization; they are prescribed weekly or monthly, making them more convenient than other osteoporosis medications. Therefore, oral bisphosphonates remain the most commonly prescribed and accessible osteoporosis treatment in NHs.¹⁴ Initiating and maintaining oral bisphosphonate therapy in NH residents may be limited by administration requirements, contraindications like dysphagia and chronic kidney disease (CKD), competing goals of care, and care transitions.^15-17 Few studies have described bisphosphonate use by NH residents,^18,19 and data used for these studies are outdated and may not be generalizable to modern NH populations.

Therefore, our aim was to describe the frequency of using drugs related to injurious falls risk (DRIFR) and oral bisphosphonates in fee-for-service Medicare beneficiaries who received NH care and had recent fracture from 2017-2022. We hypothesized that DRIFR burden would be high, but that it would be lower in residents receiving oral bisphosphonates since a provider likely reviewed their medications and prescribed osteoporosis treatment following the fracture.

Methods

Population and Data Sources

The study population included NH residents enrolled in Medicare fee-for-service and Part D from 1/1/2017-6/30/2022 who were alive and received care in the facility for 30 days. Medicare data were linked to the Minimum Data Set (MDS), Certification and Survey Provider Enhancing Reporting (CASPER), Payroll Based Journal (PBJ), and Medicare Beneficiary Summary File (MSBF). A recent fracture was identified from the active diagnosis section of the MDS and included hip and other fractures requiring evaluation, monitoring or treatment; hip and other fractures were not mutually exclusive categories.²⁰ We excluded residents who were: aged <65 years; enrolled in Medicare Advantage at index date; or without a valid MDS assessment within 3 months of the index date.

For residents who received care for more than 30 days in the same facility as of January 1, 2017, the index date was January 1, 2017. For newly admitted residents, the index date was day 31 after admission.

Study Medications and Ascertainment

Medication “use” was ascertained during the 30 days of available Medicare Part D claims before the index date and defined using Part D dispensing records. We restricted to prescription medications administered by the following routes: oral, intramuscular, subcutaneous, inhalation, or transdermal patch. For each class, medication use was defined using the days supplied regardless of dose or number of medications prescribed. The eighteen classes of DRIFR included: anticholinergics, serotonin-norepinephrine reuptake inhibitors, selective serotonin reuptake inhibitors, serotonin receptor antagonist and reuptake inhibitors (e.g., trazodone), tricyclic antidepressants, atypical antipsychotics, typical antipsychotics, barbiturates, benzodiazepines, hypnotics, loop diuretics, glucocorticoids, muscle relaxers, opioids, statins, non-selective beta-blockers, newer anticonvulsants (approved on/after 1993), and older anticonvulsants (approved before 1993). Oral bisphosphonates included alendronate, risedronate, and ibandronate.

Resident and Facility Characteristics

Patient characteristics were identified from the most recent MDS assessment (≤92 days before index date). Demographic data, including age, sex, and race/ethnicity (Non-Hispanic White, Non-Hispanic Black/African American, Hispanic, Asian/Pacific Islander, American Indian/Alaska Native, Other/Unknown), were derived from the MBSF.

Comorbid conditions were identified from the closest preceding MDS assessment to the index date and categorized as (yes/no) for an anxiety disorder, arthritis, atrial fibrillation, coronary artery disease, cerebrovascular disease or stroke, depression, dementia, diabetes, epilepsy, heart failure, hyperlipidemia, osteoporosis, Parkinson’s disease, peripheral vascular disease/peripheral artery disease, and schizophrenia. Clinical characteristics from the MDS included recent falls (yes/no), body mass index (kg/m²), and visual impairment (adequate/impaired). Functional ability was measured using activities of daily living (ADL) long form score (0-28) and ability to walk in a room (total dependence/extensive assistance/independent to limited assistance).^21,22 Hospitalizations (any cause) in the prior year were derived from MedPAR.

Facility characteristics were sourced from the CASPER and the PBJ datasets closest to the index date. These characteristics included the facility’s Five-Star rating; Medicaid enrollment percentage; staffing levels; number of available beds; and whether the facility was for-profit or in an urban location. Distance to the nearest hospital (miles) was calculated using prior methods.²³ We restricted descriptives to observations with non-missing values (<1% of observations).

This study was approved by the X and X Institutional Review Boards with a waiver of informed consent.

Statistical Analysis

Descriptive statistics were calculated using t-tests for normally distributed variables, Wilcoxon rank-sum test for non-normally distributed variables, and Chi-Square tests for categorical variables. Continuous variables were reported as means with standard deviations. Categorical variables were reported as frequencies with 95% confidence intervals (CI). Medication prevalence was reported for the study sample, with 95% CI to quantify the precision of estimates for the underlying population. Patient characteristics and medication frequency were stratified by bisphosphonate use (yes/no) to characterize DRIFR burden in the context of guideline-recommended osteoporosis treatment. Medication frequency was also stratified by dementia status, osteoporosis diagnosis, degree of functional dependence (ADL<15, ADL 15-21, ADL>21), and fracture type to identify those at greatest risk of undertreatment or high-risk prescribing. Residents with hip fracture were classified as hip fracture regardless of whether they also had other fractures; all others as other fractures. Analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC).

Results

We identified 192,118 NH residents with a recent fracture from 14,559 NHs during the study period. Of recent fractures, 41.4% were hip fractures and 67.7% were other fractures (specific types of other fractures were not available; Table 1). The mean age was 83.8 years (range=65-110 years), and 75.4% of residents were female. Residents were predominately non-Hispanic White (86.5%). Recent falls (87.1%) and comorbidities were common, with 47.4% of residents having hyperlipidemia, 46.4% depression, and 50.9% dementia diagnoses.

Table 1.

Baseline characteristics of residents overall and by bisphosphonate use, 2017-2022.

Characteristic % or mean (SD)	Full sample n=192,118	Bisphosphonate users* n=8,145	Non-users* n=183,973	P-value
Age at index† date, years	83.8 (8.8)	82.7 (8.3)	83.8 (8.7)	<0.001
Female sex	75.4	88.9	74.8	<0.001
Race/ethnicity ‡				<0.001
Non-Hispanic White	86.5	85.0	86.6
Non-Hispanic Black	5.4	3.7	5.5
Asian/Pacific Islander	2.0	3.7	2.0
Hispanic	4.8	6.3	4.8
American Indian/Alaska Native	0.5	0.8	0.5
Other/Unknown	0.7	0.6	0.8
ADL score §	17.8 (4.7)	17.0 (4.9)	17.8 (4.7)	<0.001
Recent falls (yes/no)	87.1	85.9	87.3	<0.001
Prior year any-cause hospitalization (yes/no)	45.8	39.6	46.1	<0.001
Ability to walk in a room				<0.001
Independent to limited assistance	25.8	30.2	24.6
Extensive assistance	17.0	18.2	16.9
Total dependence	57.2	51.6	57.4
History of fractures (yes/no) ║
Hip fractures	41.4	38.3	41.6	<0.001
Other fractures	67.7	72.5	67.5	<0.001
BMI, kg/m2	24.6 (6.4)	24.6 (6.3)	24.6 (6.4)	0.26
Visual impairment				<0.001
Adequate	75.3	77.6	75.2
Impaired	24.7	24.8	22.4
Comorbidities (yes/no)
Anxiety	29.1	28.4	29.2	0.13
Arthritis	30.3	34.1	30.1	<0.001
Atrial fibrillation	26.7	22.9	26.9	<0.001
CAD	23.1	19.8	23.3	<0.001
CVA/TIA/Stroke	9.5	8.3	9.6	<0.001
Depression	46.4	49.5	46.2	<0.001
Diabetes	28.4	24.7	28.6	<0.001
Epilepsy	7.0	8.5	6.9	<0.001
Heart failure	21.6	17.5	21.8	<0.001
Hyperlipidemia	47.4	51.8	47.2	<0.001
Osteoporosis	18.8	57.0	17.1	<0.001
Parkinson’s disease	5.8	5.7	5.9	0.57
PVD/PAD	8.2	7.1	8.3	<0.001
Schizophrenia	3.7	4.2	3.6	0.006
Dementia	50.9	48.3	51.0	<0.001
Nursing Home Facility Characteristics
Five Star rating	3.3 (1.4)	3.3 (1.4)	3.3 (1.4)	0.81
RN hours per resident-day	0.43 (0.28)	0.41 (0.27)	0.43 (0.28)	<0.001
CNA hours per resident-day	2.2 (0.5)	2.2 (0.5)	2.2 (0.5)	0.63
Miles to nearest hospital	3.4 (11.1)	3.6 (17.1)	3.4 (10.8)	0.50
Number of beds/facilities	134.3 (79.8)	132.9 (81.3)	134.3 (79.7)	0.12
% of residents on Medicaid	58.7 (20.5)	58.6 (20.5)	59.7 (20.1)	<0.001
For-profit	67.8	68.3	68.3	0.33
Urban location	72.5	71.1	71.1	0.005

Notes: Abbreviations: ADL, activities of daily living; BMI, body mass index; CAD, coronary artery disease; CNA, certified nursing assistant; CVA, cerebrovascular accident; PVD, peripheral vascular disease; PAD, peripheral artery disease; RN, registered nurse; SD, standard deviation; TIA, transient ischemic attack. All missing <1% of observations.

^*Medication “use” was defined using Part D dispensing records.

^†For residents who had received care for more than 30 days in the same facility as of January 1, 2017, the index date was January 1, 2017. For newly admitted residents, the index date was day 31 after admission.

^‡Race and ethnicity categories were measured using the Minimum Dataset and are not mutually exclusive. Other race/ethnicity was defined as not identifying with the specified categories.

^§Scores range from 0-28, with higher scores indicating greater impairment.

^║Hip and other fractures were not mutually exclusive categories

Only 4.2% (95% CI:4.2-4.3) of residents used oral bisphosphonates. Compared to non-users, bisphosphonate users were younger, more often female, Asian/Pacific Islander, Hispanic, American Indian/Alaska Native, and functionally independent, and had fewer hospitalizations in the prior year and recent falls. Osteoporosis diagnosis was common among residents with bisphosphonate use (Table 1).

Overall, 72.8% of residents were using one or more DRIFRs. The most frequently used DRIFRs were statins (31.1%), opioids (26.7%), newer anticonvulsants (22.1%), and loop diuretics (22.1%). Use of typical antipsychotics (0.9%), hypnotics (1.1%), barbiturates (1.3%), and tricyclic antidepressants (1.3%) were least common. Of note, 5.4% of residents used anticholinergics (5.4%), 7.7% SARIs (e.g., trazodone), and 11.9% atypical antipsychotics (Table 2). Excluding statins and non-selective beta blockers, 67.1% of residents were using one or more DRIFRs.

Table 2.

Prescribed medication use overall and concurrent with bisphosphonate use (yes/no).

Medication Classes	Overall use, % (95% CI)* (N=192,118)	Bisphosphonate use, %*† (N=8,145)	Non-users, %*† (N=183,973)	P-value
Anticholinergics	5.42 (5.32-5.52)	8.2	5.3	<0.001
Atypical Antipsychotics	11.85 (11.70-11.99)	11.6	11.9	0.51
Barbiturates	1.26 (1.22-1.32)	1.8	1.2	<0.001
Benzodiazepines	12.40 (12.26-12.55)	12.2	12.4	0.51
Glucocorticoids	5.20 (5.10-5.30)	6.9	5.1	<0.001
Hypnotics	1.09 (1.05-1.14)	1.6	1.1	0.023
Loop Diuretics	22.05 (21.87-22.24)	21.7	22.1	0.42
Muscle Relaxers	3.98 (3.90-4.07)	5.7	3.9	<0.001
Newer Anticonvulsants ‡	22.05 (21.87-22.24)	25.3	21.9	<0.001
Non-Selective Beta-Blockers	8.70 (8.57-8.82)	8.08	8.72	0.043
Older Anticonvulsants §	1.56 (1.51-1.62)	2.4	1.5	<0.001
Opioids	26.72 (26.52-26.92)	29.5	26.6	<0.001
SARI	7.69 (7.57-7.81)	7.4	7.7	0.34
SNRI	6.87 (6.75-6.98)	9.0	6.8	<0.001
SSRI	20.13 (19.95-20.31)	21.7	20.1	<0.001
Statins	31.06 (30.86-31.27)	37.5	30.8	<0.001
Tricyclic Antidepressants	1.32 (1.27-1.37)	1.7	1.3	0.003
Typical Antipsychotics	0.93 (0.88-0.97)	1.0	0.9	0.76
Overall FRI-Associated Classes	72.81 (72.61-73.01)	76.6	72.7	<0.001
Bisphosphonates	4.24 (4.15-4.33)	-	-

Notes: Abbreviations: CI, confidence interval; FRI, fall-related injury; SARI, Serotonin Receptor Antagonists and Reuptake Inhibitor (e.g., trazodone); SNRI, Serotonin-Norepinephrine Reuptake Inhibitor; SSRI, Selective Serotonin Reuptake Inhibitor.

^*Percent is estimated prevalence; 95% CI quantifies precision for underlying population.

^†Medication “use” was defined using Part D dispensing records.

^‡Newer anticonvulsants were those approved on/after 1993.

^§Older anticonvulsants were those approved before 1993.

The mean number of DRIFRs was 1.9 (SD=1.7), with a median of 2 (IQR=0-3). While 27.2% of residents did not use any DRIFRs, 33.4% of residents used three or more DRIFR classes. Residents with bisphosphonate use had higher mean use of DRIFR classes compared to non-users (2.1 vs. 1.9, p=<0.001). Residents with bisphosphonate use were more likely to use glucocorticoids, SSRIs, and both newer and older anticonvulsants, among others (Table 2).

Residents with dementia were more likely to use SSRIs, SARIs, antipsychotics, and benzodiazepines (Supplemental Table 1). Residents with osteoporosis were more likely to use anticholinergics, glucocorticoids, opioids, and bisphosphonates (Supplemental Table 2). Residents with severe ADL limitations (ADL>21) were more likely to use antipsychotics, with lower use of other DRIFRs and bisphosphonates (Supplemental Table 3). In the hip fracture subgroup, residents were more likely to use atypical antipsychotics, with lower use of other DRIFRs and bisphosphonates (Supplemental Table 4).

Discussion

From our large sample of NH residents with recent fractures, we found that DRIFR use was common, whereas oral bisphosphonate use was rare. Many residents used one or more DRIFR, including 76.6% of residents with oral bisphosphonate use. Given the high risk of recurrent injury in NH residents with fracture,²⁴ our results suggest that there may be potential opportunities to improve secondary fracture prevention efforts not only with osteoporosis treatment but also through deprescribing efforts.

Little literature exists documenting the prevalence of bisphosphonate and FRI-associated medication class use in NH residents with fracture. One study (2015) of female French NH residents documented low rates of bisphosphonate use after fracture (10.3%).²⁵ However, prescribing patterns may differ between France and the US, and this study had a low sample size (n=116).²⁵ An older US-based NH study (1999-2004) of New Jersey Medicaid beneficiaries found 5.5% of residents with fracture received osteoporosis treatment (bisphosphonates (intravenous and oral), raloxifene, teriparatide, calcitonin, and hormone replacement therapy).¹⁹ Another US-based NH study (2003-2004) among residents 50 years or older with recent hip fracture or diagnosed osteoporosis from two states found 19% of residents were prescribed bisphosphonates.¹⁸ Higher rates of bisphosphonate treatment may be due to differences in the populations since women are more likely to receive osteoporosis treatment,¹⁸ and those with a diagnosis of osteoporosis are more likely to receive treatment. Moreover, trends in bisphosphonate use have declined due to safety announcements,²⁶ likely contributing to the observed low frequency of use.

Our findings highlight the need for medication optimization as a core component of post-fracture care. Though non-oral osteoporosis treatments are available, evidence for their use in NHs is limited; oral bisphosphonates are recommended as the first-line treatment according to the Society for Post-Acute and Long-Term Care Medicine.^14,17 While bisphosphonates are appropriately prescribed to reduce future fracture risk by improving bone health, concurrent use of DRIFRs may offset the net benefit of bisphosphonates by increasing fall and injury risk, which motivated our stratification of analyses by bisphosphonate use.^17,27 Contrary to our expectation, DRIFR burden was higher among bisphosphonate users than non-users, which may reflect intentional fracture risk reduction efforts, greater multimorbidity, post-fracture pain management, glucocorticoid-associated osteoporosis management, or greater overall medication burden in post-fracture residents. Glucocorticoid use, for instance, was more common in residents with bisphosphonate treatment, which is appropriate given the adverse effects of glucocorticoids on bone. Importantly, low oral bisphosphonate use may not represent a universal care gap, as contraindications, personal preferences, and competing health priorities may appropriately preclude bisphosphonate use. Efforts to prevent secondary fractures may benefit from a medication review that includes consideration of osteoporosis treatment and deprescribing drugs associated with falls and fractures.

When considering deprescribing targets, some medication classes may confer greater risk of injury and/or more opportunities for reduction than others. Moreover, the strength of evidence implicating medication classes in FRIs varies, as does whether deprescribing these medications reduces FRI risk. Additionally, some DRIFRs may remain clinically appropriate despite their associated risk. Psychotropic medications are often the focus of deprescribing efforts in NH residents, but other medication classes may warrant attention.^28,29 While opioids and newer anticonvulsants (e.g., gabapentin) may be prescribed appropriately for pain following fracture, these classes have been associated with increased fall and fracture risk and were common in this population. Our results suggest these medications may represent common candidates for medication review. However, differences in prescribing patterns in residents with dementia, an osteoporosis diagnosis, functional dependence, and across fracture types suggest that medication-related FRI risk and potential undertreatment may vary substantially across NH residents, underscoring the need for more individualized prescribing approaches. For example, lower use of newer anticonvulsants and opioids among residents with dementia may reflect intentional avoidance of medication-related adverse events or inadequate attention to pain management in cognitively impaired residents. Future work should assess the impact of optimizing use of these medications, including deprescribing when appropriate, on subsequent fractures and functional outcomes.

We acknowledge several limitations of our study. First, the sample excluded Medicare Advantage beneficiaries, who typically fill fewer high-risk medications than fee-for-service beneficiaries possibly due to aggressive utilization management strategies in Medicare Advantage plans.³⁰ Second, while our 30-day window would be expected to capture most oral bisphosphonate prescriptions, residents prescribed oral bisphosphonates through multi-month dispensings before the lookback period may have been misclassified as non-users. Because medications in NHs are typically dispensed monthly, we expect the magnitude of this misclassification to be limited. Third, we did not assess exposure to multiple medications within the same class. Fourth, we were unable to further categorize non-hip fractures. Since fracture severity and clinical management strategies may differ across fracture types, this may limit the generalizability of our findings across fracture types. Lastly, we only considered oral bisphosphonate use for osteoporosis treatment; overall osteoporosis treatment prevalence may therefore be underestimated, as alternative therapies may be preferred in residents with dysphagia, severe CKD, or other contraindications. However, previous research has noted low use of alternative agents.¹⁸ Additionally, oral bisphosphonate use may be challenging due to administration requirements, competing health or medication priorities, and care transitions.

Despite these limitations, our population included a nationally representative sample of NH residents at high-risk for FRIs with attention to multiple medication classes. Additionally, we included many classes of DRIFRs, including non-sedating medications like statins and non-selective beta blockers. Exclusion of these medications reduced overall DRIFR prevalence by 5.7 percentage points, but the prevalence remained high at 67.1%.

Conclusions and Implications

In summary, older adults with fracture are vulnerable to subsequent falls and fractures, a vulnerability that may be increased by high DRIFR burden and low frequency of osteoporosis medication use. Comprehensive medication review should be a standard component of post-fracture management in NHs to minimize further injuries.

Data Statement:

Research code available upon request. For Medicare claims data, contact the Research Data Assistance Center (ResDAC).