Abstract
Purpose:
Several non-oral drugs exist for osteoporosis treatment, including zoledronic acid (ZA), denosumab, and teriparatide. Little data exist on the comparative effectiveness of these drugs for hip fracture prevention in frail older adults. We examined their comparative effectiveness in one of the frailest segments of the US population—nursing home (NH) residents.
Methods:
We conducted a national retrospective cohort study of NH residents aged ≥65 years using 2012 to 2016 national US Minimum Data Set clinical assessment data and linked Medicare claims. New parenteral ZA, denosumab, and teriparatide use was assessed via Medicare Parts B and D; hip fracture outcomes via Part A; and 125 covariates for confounding adjustment via several datasets. We used inverse probability weighted (IPW) competing risk regression models to compare hip fracture risk between groups with teriparatide as the reference.
Results:
The study cohort (N=2,019) included 1,046 denosumab, 578 teriparatide, and 395 ZA initiators. Mean age was 85 years, 90% were female, and 68% had at least moderate functional impairment. 72 residents (3.6%) had a hip fracture and 1,100 (54.5%) died over a mean follow-up of 1.5 years. Compared to teriparatide use, denosumab use was associated with a 46% lower risk of hip fracture (HR 0.54, 95%CI 0.29–1.00) and no difference was observed for ZA (HR 0.70, 95%CI 0.26–1.85).
Conclusions:
Denosumab and ZA may be as effective as teriparatide for hip fracture prevention in frail older adults. Given their lower cost and easier administration, denosumab and ZA are likely preferable non-oral treatments for most frail, older adults.
Keywords: Hip Fractures, Nursing Homes, Denosumab, Teriparatide, Zoledronic Acid, Comparative Effectiveness Research
MINI ABSTRACT
The comparative effects of zoledronic acid, denosumab, and teriparatide for preventing hip fractures in frail older adults, especially those in nursing homes, was unknown. We found that denosumab and zoledronic acid may be as effective as teriparatide for hip fracture prevention in nursing home residents.
INTRODUCTION
Nursing home (NH) residents are one of the oldest and frailest segments of the U.S. population. As such, over four out of five NH residents have osteoporosis and rates of hip fracture are nearly 10-fold greater than rates in non-institutionalized persons.1,2 Hip fractures have devastating consequences for NH residents, including mortality, frequent rehospitalization, and functional decline.1,3–5 Data specific to this population suggest that oral bisphosphonates are an effective first-line treatment option for the prevention of hip fracture, capable of reducing the risk by an average of 17%.6 Other osteoporosis treatments, such as parenteral bisphosphonates, denosumab, and teriparatide, may be attractive alternative treatment options among frail older adults based on their potency or anabolic potential.7–9 Among community dwellers, parenteral bisphosphonates and denosumab have been found to reduce the risk of hip fracture compared with placebo, whereas the effect of teriparatide on hip fracture risk specifically remains uncertain.10,11 Nonetheless, these treatments have the advantage of avoiding increased pill burden, improving adherence, and averting some common adverse drug effects like dysphagia and esophagitis.12
Despite the substantial burden of hip fracture and need for osteoporosis treatment among NH residents and other frail older adults, very limited data exist on the efficacy of these newer agents in this population. Only one randomized clinical trial (RCT) of osteoporosis treatments has included NH residents or other frail older adults.13 Even among healthier populations, there are few RCTs that directly compare the efficacy of alternative osteoporosis treatments, and most RCTs are not sufficiently powered to detect a difference in hip fracture outcomes between treatment groups.14–18 A prime example is the VERtebral fracture treatment comparisons in Osteoporotic women (VERO) study.14 This trial of postmenopausal women (mean age 72.1 years) was powered to detect a difference in overall fracture risk between the teriparatide and risedronate groups, but it was insufficiently powered to detect a difference in hip fracture risk.14 Evidence-based prescribing and the clinical management of osteoporosis will continue to be a major challenge without such information. Observational studies with larger sample sizes and more hip fracture events are one practical and efficient solution to generate needed evidence on how osteoporosis treatments compare for fracture prevention, especially among older adults in the NH and other real-world settings.
To address the lack of comparative effectiveness data for NH residents and other frail older adults, we examined the comparative effectiveness of zoledronic acid (ZA), denosumab, and teriparatide for hip fracture prevention among older NH residents. We hypothesized that there would be notable reductions in hip fracture risk for ZA and denosumab users compared to teriparatide users.
METHODS
Study Design and Data Sources
This was a retrospective new-user cohort study that linked the following national datasets: Medicare fee-for-service enrollment information, Medicare Part A inpatient hospital claims, Medicare Part B outpatient claims, Medicare Part D prescription drug claims, and Minimum Data Set (MDS) version 3.0.19 The MDS is a comprehensive, clinical assessment instrument used to document health status of NH residents, including demographic, medical, functional status, psychological, and cognitive status information. The MDS assessments are federally mandated for all residents in NHs certified to receive Medicare or Medicaid funding. Certification And Survey Provider Enhanced Reports (CASPER) data were used for facility-level information, including NH characteristics, staffing levels, and quality measures. A previously validated algorithm was used to track the timing and location of health service use.20 Our observational study was designed to emulate a hypothetical multi-arm pragmatic RCT that could have been conducted had it been feasible.21,22 This study was approved by the Hebrew SeniorLife Institutional Review Board.
Study Population
The study population was adults aged ≥65 years who became long-stay NH resident (>100 days in the NH) between January 1, 2012 and December 31, 2012. The index date was the first eligible dispensing of ZA, denosumab, or teriparatide between January 1, 2013 and December 31, 2015 after becoming a long-stay resident. Individuals with a dispensing of any of the three osteoporosis drug classes of interest in the 365-days before the index date were excluded. Study subjects were excluded if they had less than 365 days of continuous enrollment in fee-for-service Medicare Parts A, B, and D, were enrolled in Medicare Advantage at any time, or had evidence of probable cancer documented one year prior to or at the time of the first eligible drug dispensing. Individuals were identified as having probable cancer in one of three ways: 1) if there was direct documentation of any cancer, 2) if they had chemotherapy use, or 3) if the dose or frequency of ZA or denosumab was consistent with that for cancer-related indications. Direct documentation of cancer was obtained using Section I of the MDS or any ICD code (detailed in a prior publication23) for solid or hematologic cancers in the Part A inpatient claims. Chemotherapy was identified using Section O of the MDS and previously enumerated claims codes.24 Dosing related exclusions included 1) one or more four milligram (mg) doses of ZA or 14–300 day intervals between doses (osteoporosis dosing is typically 5 mg not more than once per year), or 2) one or more 120 mg doses of denosumab or 14–119 day intervals between doses (osteoporosis dosing is typically 60 mg every 6 months). Other exclusions are shown in Supplementary Table 1.
Exposure
Following the approach used by previous investigators, initiation of denosumab, ZA, or teriparatide were assessed using both Medicare Part B and Part D claims.25,26 We used Healthcare Common Procedure Coding System (HCPCS) J codes to identify treatment use in Part B and national drug codes to identify use in Part D. New use was defined as the first dispensing of one of the treatments after 365 days without a dispensing of any of the three drug classes of interest. It is worth noting that in the initial design of the study, we had considered including ibandronate and pamidronate in the exposure definition, but preliminary analyses of the study data demonstrated that there were too few users of each and reporting on them would have violated CMS’s Cell Size Suppression Policy governing our use of the data.
Causal Contrasts
The primary causal contrasts of interest were defined as the effects of initiating denosumab versus teriparatide or ZA versus teriparatide regardless of subsequent treatment discontinuation or switching among treatment groups (i.e., the observational study analog of the intention-to-treat estimand).21,22 We chose teriparatide as the reference group because it is associated with a greater cost and burden of administration than ZA or denosumab.
Outcome
The sole outcome was hospitalized hip fracture defined using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9) code 820.xx and ICD-10 code S72.xx in the primary or secondary position on inpatient claims.27,28 In prior validation studies, the positive predictive value (PPV) was 98% for the ICD-9 version of the hip fracture definition.28
Follow-Up
Follow-up started on the day of the first ZA, denosumab, or teriparatide dispensing between January 1, 2013 and December 31, 2015. Follow-up continued until Medicare disenrollment (from Parts A, B, or D) or enrollment in Medicare Advantage, death, hip fracture, or study end (December 31, 2016).
Baseline Characteristics
Characteristics that could potentially confound the relationship between the receiving denosumab, ZA, or teriparatide and the hip fracture outcome were prespecified and all measured before the index date.6,29,30 Characteristics were measured up to 365 days prior to the index date, with the exception of prior length of NH stay. A complete list of these 125 characteristics and details about their measurement are provided in Supplementary Table 2.
Statistical Analyses
We adjusted for confounding by baseline covariates using methods that rely on estimating the propensity score (i.e., the joint probabilities of receiving denosumab versus teriparatide and ZA versus teriparatide, conditional on covariates). We estimated the propensity score via a multinomial logistic regression model that used the aforementioned 125 baseline variables to predict the use of each of the three medication exposures of interest. The propensity scores were used to construct stabilized inverse probability of treatment weights (IPTWs), which resulted in good covariate balance across treatment groups based on standardized mean differences in covariates (Supplementary Table 3).
We used IPTW linear regression models fit with ordinary least squares to estimate risk differences (RDs) with non-parametric bootstrap 95% confidence intervals (CIs). Bootstrap CIs were calculated using 10,000 replicates. We used IPTW Fine and Gray competing risk regression models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) comparing denosumab versus teriparatide and ZA versus teriparatide for the outcome of hip fracture while accounting for the competing risk of death. To accompany the Fine and Gray estimates, we plotted the IPTW cumulative incidences of hip fracture by treatment.
Stability and Sensitivity Analyses
We conducted several stability analyses to test the robustness of our treatment effect estimates to analytic decisions. In the first and second stability analyses, we used standard unweighted and IPTW Cox proportional hazards regression models as an alternative to the primary models accounting for the competing risk of death. In the third and fourth stability analyses, we estimated the propensity scores using generalized boosted regression models and then re-estimated the effect of treatment on outcomes using an IPTW Cox regression model and an IPTW Fine and Gray competing risk regression model. In a fifth stability analysis, we estimated the effect of adhering to the treatment initiated (being “on treatment”; the per-protocol estimand) in unweighted, IPTW, and joint IPTW and inverse probability of censoring (IPCW) by treatment discontinuation-weighted analyses. In this fifth analysis, individuals were censored at their first discontinuation of one of the treatments. Discontinuation was defined as a maximum of the following number of days since the last observed dispensing without a subsequent dispensing: 182 days for denosumab, 365 days for ZA, or 30 days for teriparatide. Of note, discontinuations due to switches from one treatment to another were rare enough that reporting the incidence of each would violate CMS’s Cell Size Suppression Policy governing our use of the study data. Finally, in a sixth analysis, we modified the fifth analysis to allow for gaps between prior and subsequent dispensings of the drug. These gaps were 365 days for ZA, 182 days for denosumab, and 60 days for teriparatide.31 As an example, an individual who had a dispensing of ZA and then did not receive their second dispensing of ZA until two years later would still be considered exposed up to the time of the second dispensing (i.e., 365 days of exposure + 365 days in the gap period = 720 days of continuous exposure before the second dispensing). Finally, we conducted a sensitivity analysis using the E-value (see Supplementary Methods 1).
Software
Data were analyzed using SAS, version 9.4 (SAS Institute, Inc, Cary, NC), Stata, version 16.0 (Stata Corp, College Station, TX), and R version 3.6.2 software.
RESULTS
Study Cohort
The study cohort (N=2,019) included 1,046 denosumab, 578 teriparatide, and 395 ZA initiators (Table 1). Mean age was 85 years (SD 7.9), median (IQR) was 86 (79–90), and the range 66 to 107. The study population was 90% female, and 68% had at least moderate functional impairment. Alzheimer’s or non-Alzheimer’s dementia was documented for 62% of the cohort. Over 77% of the cohort had at least one Beers Criteria32 potentially inappropriate medication and the average number of prescription medications dispensed was 16 (SD 7) in the year prior to the index date. Individuals resided in the NH a median of 3.2 years (IQR 2.0–5.2) prior to osteoporosis treatment initiation.
Table 1.
Baseline characteristics among frail older adults initiating non-oral treatments for osteoporosis while in the nursing home (prior to inverse probability of treatment weighting), Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012 to 2016 (N=2,019).
| Teriparatide | Zoledronic Acid | Denosumab | |
|---|---|---|---|
| Characteristic | (n=578) | (n=395) | (n=l,046) |
| Age years, mean (SD) | 84.2 (8.1) | 84.4 (8.0) | 85.3 (7.7) |
| Female sex | 518 (89.6%) | 329 (83.3%) | 967 (92.4%) |
| Race | |||
| White, non-Hispanic | 463 (80.1%) | 301 (76.2%) | 855 (81.7%) |
| Black, non-Hispanic | 41 (7.1%) | 49 (12.4%) | 87 (8.3%) |
| Hispanic | 51 (8.8%) | 33 (8.4%) | 60 (5.7%) |
| Other | 23 (4.0%) | 12 (3.0%) | 44 (4.2%) |
| Body mass index, kg/m2, mean (SD) | 25.7 (6.1) | 26.1 (6.1) | 26.0 (6.3) |
| Cognitive Function Scale | |||
| Intact Cognition | 185 (32.0%) | 114 (28.9%) | 368 (35.2%) |
| Mild Impairment | 129 (22.3%) | 87 (22.0%) | 209 (20.0%) |
| Moderate Impairment | 202 (34.9%) | 127 (32.2%) | 364 (34.8%) |
| Severe Impairment | 62 (10.7%) | 67 (17.0%) | 105 (10.0%) |
| Activities of Daily Living 28-point Long Form Summary Scale, median (IQR) | 18 (13, 21) | 19 (14, 22) | 18 (11, 20) |
| Transfer performance | |||
| Independent | 66 (11.4%) | 31 (7.8%) | 148 (14.1%) |
| Supervision needed | 48 (8.3%) | 42 (10.6%) | 104 (9.9%) |
| Limited assistance needed | 78 (13.5%) | 58 (14.7%) | 149 (14.2%) |
| Extensive assistance needed | 288 (49.8%) | 173 (43.8%) | 491 (46.9%) |
| Total dependence | 98 (17.0%) | 91 (23.0%) | 154 (14.7%) |
| Fall in prior six months | 118 (20.4%) | 33 (8.4%) | 102 (9.8%) |
| No. of medications, mean (SD) | 16.8 (7.6) | 13.4 (6.3) | 16.5 (7.0) |
| 2015 Beers List potentially inappropriate medication use | 473 (81.8%) | 260 (65.8%) | 826 (79.0%) |
| Prior osteoporosis medication use | |||
| Selective estrogen receptor modulator | 17 (2.9%) | 13 (3.3%) | 42 (4.0%) |
| Estrogen hormone replacement | 26 (4.5%) | 19 (4.8%) | 48 (4.6%) |
| Calcitonin intranasal spray | 41 (7.1%) | 23 (5.8%) | 85 (8.1%) |
| Oral bisphosphonate | 120 (20.8%) | 53 (13.4%) | 357 (34.1%) |
| Charlson Comorbidity Index | |||
| 1 | 140 (24.2%) | 87 (22.0%) | 217 (20.7%) |
| 2–3 | 260 (45.0%) | 183 (46.3%) | 445 (42.5%) |
| >=3 | 178 (30.8%) | 125 (31.6%) | 384 (36.7%) |
| Medical conditions | |||
| Atrial fibrillation or other dysrhythmia | 98 (17.0%) | 62 (15.7%) | 177 (16.9%) |
| Coronary artery disease | 98 (17.0%) | 87 (22.0%) | 223 (21.3%) |
| Renal insufficiency, renal failure, or end-stage renal disease | 35 (6.1%) | 17 (4.3%) | 116 (11.1%) |
| Diabetes | 179 (31.0%) | 104 (26.3%) | 297 (28.4%) |
| Osteoarthritis or rheumatoid arthritis | 194 (33.6%) | 147 (37.2%) | 451 (43.1%) |
| Hip fracture, prior | 38 (6.6%) | 13 (3.3%) | 39 (3.7%) |
| Fracture at location other than hip, prior | 77 (13.3%) | 11 (2.8%) | 64 (6.1%) |
| Alzheimer’s Disease or related dementias | 347 (60.0%) | 260 (65.8%) | 644 (61.6%) |
| Cerebrovascular accident, transient ischemic attack, or stroke | 79 (13.7%) | 71 (18.0%) | 148 (14.1%) |
| Parkinson’s Disease | 43 (7.4%) | 28 (7.1%) | 86 (8.2%) |
| No. of hospitalizations | |||
| 0 | 359 (62.1%) | 306 (77.5%) | 799 (76.4%) |
| 1 | 154 (26.6%) | 60 (15.2%) | 179 (17.1%) |
| >=2 | 65 (11.2%) | 29 (7.3%) | 68 (6.5%) |
| Length of nursing home stay, years, median (IQR) | 2.9 (1.9, 4.7) | 3.4 (2.0, 5.5) | 3.2 (2.0, 5.2) |
| For-profit nursing home | 467 (80.8%) | 316 (80.0%) | 769 (73.5%) |
Prior to IPTW, there were a number of notable differences in baseline characteristics between treatment groups (Table 1; Supplementary Table 3). Teriparatide and denosumab initiators were more likely to have received an oral bisphosphonate or a Beers Criteria potentially inappropriate medication. ZA and denosumab initiators were less likely to have a history of hip fracture or a recent fall. Teriparatide and ZA initiators were more likely to reside in for-profit NHs.
During an average follow-up of 1.5 years, 72 residents (3.6%) had a hip fracture and 1,100 (54.5%) died. There were 190 deaths (48.1%) among ZA users, 310 deaths (53.6%) among teriparatide users, and 600 deaths (57.4%) among denosumab users, corresponding to mortality rates of 295, 351, and 399 deaths per 1,000 person-years, respectively.
Treatment Effects
The crude incidence rates of hip fracture per 1,000 person-years were 29.4 (95%CI 20.0–43.2) for teriparatide initiators, 23.3 (95%CI 14.1–38.7) for ZA initiators, and 20.1 (95%CI 14.5–29.3) for denosumab initiators (Table 2). Compared to teriparatide, denosumab was associated with a 46% lower risk of hip fracture (IPTW HR 0.54, 95%CI 0.29–1.00). No difference was observed for ZA (HR 0.70, 95%CI 0.26–1.85). The corresponding absolute risk difference estimates were −1.71% (95%CI −5.12, 1.70) for ZA and −2.04% (95%CI −4.40, 0.32) for denosumab (Table 3; Figure 1).
Table 2.
Relative treatment effect estimates (intention-to-treat) for hip fracture outcome among frail older adults initiating non-oral osteoporosis treatments, Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012–2016 (N=2,019).
| Treatment | Number of initiators | Number of hip fractures | Person-years | Incidence rate (95% CI) (per 1,000 person-years) | Crude HR (95% CI) | Adjusted (IPTW) HR (95% CI) |
|---|---|---|---|---|---|---|
| Teriparatide | 578 | 26 | 883 | 29.4 (20.0–43.2) | 1.00 (reference) | 1.00 (reference) |
| Zoledronic Acid | 395 | 15 | 644 | 23.3 (14.1–38.7) | 0.83 (0.44–1.57) | 0.70 (0.26–1.85) |
| Denosumab | 1,046 | 31 | 1,506 | 20.1 (14.5–29.3) | 0.67 (0.40–1.12) | 0.54 (0.29–1.00) |
Abbreviations: CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighted.
Table 3.
Main absolute treatment effect estimates (intention-to-treat) for hip fracture outcome among frail older adults initiating non-oral osteoporosis treatments, Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012 to 2016 (N=2,019).
| Exposure | Number of initiators | Number of hip fractures | Cumulative Incidence, Percentage | Crude Risk Difference (95% CI), Percentage | Adjusted (IPTW) Risk Difference (95% CI), Percentage |
|---|---|---|---|---|---|
| Teriparatide | 578 | 26 | 4.50 (2.81, 6.19) | 0.00 (reference) | 0.00 (reference) |
| Zoledronic Acid | 395 | 15 | 3.80 (1.91, 5.68) | −0.70 (−1.83, 3.23) | −1.71 (−5.12, 1.70) |
| Denosumab | 1,046 | 31 | 2.96 (1.94, 3.99) | −1.53 (−3.51, 0.44) | −2.04 (−4.40, 0.32) |
Abbreviations: CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighted.
Figure 1.
Cumulative incidence function plot of hip fracture outcome for initiators of non-oral osteoporosis treatments after inverse probability of treatment weighting, Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012 to 2016 (N=2,019).
Stability and Sensitivity Analyses
The per-protocol estimates of treatment effect were consistent with the main intention-to-treat results (Table 4). After applying IPTW and ICPW, denosumab use was associated with a 52% lower risk of hip fracture (HR 0.48, 95%CI 0.26–0.88) and no difference was observed for ZA use (HR 1.02, 95%CI 0.31–3.31) compared with teriparatide use. After permitting gaps between drug dispensings, the as-treated results were still generally concordant (Supplementary Table 4). Results from other stability analyses (Figure 2) were generally consistent for denosumab versus teriparatide, with HRs ranging from 0.54 (95%CI 0.30–0.99) to 0.69 (95%CI 0.41–1.15). The results for ZA were more sensitive to the analytic approach, with HRs ranging from 0.55 (95%CI 0.20–1.53) to 1.48 (95%CI 0.53–4.17) depending on the approach employed (Supplementary Table 4; Figure 2). Results were modestly robust to confounding based on the E-values (Supplementary Table 5).
Table 4.
Stability analysis (per-protocol) relative treatment effect estimates for hip fracture outcome among frail older adults initiating non-oral osteoporosis treatments, Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012 to 2016 (N=2,019).
| Exposure | Number of initiators | Number of hip fractures | Person-years | Incidence rate (95% CI) (per 1,000 person-years) | Crude HR (95% CI) | Adjusted (IPTW) HR (95% CI) | Adjusted (IPTW & IPCW) HR (95% CI) |
|---|---|---|---|---|---|---|---|
| Teriparatide | 578 | 26 | 135 | 193.1 (131.5–283.6) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
| Zoledronic Acid | 395 | 15 | 486 | 30.9 (18.6–51.2) | 0.58 (0.31–1.07) | 0.50 (0.17–1.46) | 1.02 (0.31–3.31) |
| Denosumab | 1,046 | 31 | 1,128 | 27.5 (19.3–39.1) | 0.45 (0.28–0.73) | 0.39 (0.22–0.70) | 0.48 (0.26–0.88) |
Abbreviations: CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighted; ICPW, inverse probability of censoring by treatment discontinuation weighted.
Figure 2. Alternative analyses (stability analyses) testing the robustness of the main results to various decisions about propensity score and outcome estimation, Medicare linked to Minimum Data Set and Certification And Survey Provider Enhanced Reports data from 2012 to 2016 (N=2,019).
Teriparatide is the reference and hip fracture is the outcome for all analyses. Abbreviations: Unweighted, no use of inverse probability weights to account for measured confounding; Cox, Cox proportional hazards regression outcome model used for the outcome estimation; IPW, outcome model weighted using inverse probability of treatment weights to adjust for measured confounding between the treatment groups; GBM, generalized boosted regression models used to estimate the propensity scores and construct the inverse probability of treatment weights; F&G, Fine and Gray competing risk regression model used for the outcome analysis.
DISCUSSION
In this national retrospective cohort study, we found that ZA and denosumab appeared to be as effective as teriparatide for hip fracture prevention in older, predominantly frail adults residing in NHs. Most analyses suggested that denosumab was associated with a significant reduction in hip fracture risk compared to teriparatide, which was concordant with our a priori hypothesis. Given the annual dosing of ZA and biannual dosing of denosumab in comparison with the daily dosing of teriparatide, these agents are likely easier for NH staff to administer to residents. In light of their seemingly comparable effectiveness and much lower cost than teriparatide, ZA and denosumab might be preferred treatment options for many frail older adults residing in NHs.
There have been very few prior direct comparisons of denosumab, ZA, and teriparatide. A majority of studies have been underpowered to detect fracture outcomes and have instead focused on bone mineral density.14,33,34 Even a 2019 network meta-analysis of available trials reporting hip fracture outcomes produced imprecise estimates.35 The meta-analysis reported a risk ratio (RR) of 0.88 (95%CI 0.30–2.56) for denosumab versus teriparatide and 0.94 (95%CI 0.35–2.50) for ZA versus teriparatide.35 Most prior trials of osteoporosis treatments have excluded long-term NH residents because of difficulties recruiting them and their high prevalence of characteristics like very advanced age, multimorbidity, polypharmacy, and frailty, which is the decreased ability of individuals to recover from physiological insults.36–38 Frailty is particularly important because it can modify the effects of preventive pharmacologic treatments, resulting in a reduced benefit of treatment, new or more pronounced potential harms, and an entirely different benefit-harm ratio than that observed for the younger and healthier populations that are ultimately included in RCTs. Thus, it is difficult to directly compare our findings for hip fracture outcomes in frail older adults residing in NHs.
It remains generally difficult to assess the concordance of our findings with the totality of prior data because fracture events are rare and prior comparative effectiveness estimates are compatible with an uninformatively wide range of interpretations, including that teriparatide is as effective, more effective, and less effective for hip fracture prevention than denosumab or ZA.9,39,40 To illustrate, the authors of a recent meta-analysis reported a pooled odds ratio of 0.54 with a 95%CI of 0.25–1.17 (p=0.12) when synthesizing studies comparing teriparatide to active comparators for hip fracture prevention.40 Given that the CI contains the null of 1.00 and has an upper bound of 1.17, the data are compatible with the conclusions that teriparatide is no more effective than active comparators and may even increase the odds of hip fracture by up to 17% in comparison. The data are also compatible with the conclusion that teriparatide decreases the odds of hip fracture by 46% and perhaps by as much as up to 75% given the lower CI bound of 0.25.
In the absence of a clear benefit of denosumab or ZA over teriparatide, the cost implications must be considered. Denosumab and teriparatide are both significantly more expensive than ZA. Denosumab is at least twice as expensive and teriparatide is at least four times as expensive as ZA.41 When overlaid with the more frequent daily dosing of teriparatide, denosumab and ZA are more cost-efficient and convenient treatment options for many frail older adults and the staff who care for them in NHs. Indeed, the notable reduction in person-time for teriparatide from the ITT analysis (883 person-years) to the per-protocol analysis (135 person-years) may suggest that NH staff burden and NH resident pain or discomfort from daily injections may present major challenges to treatment persistence. In contrast, promoting the use of denosumab and ZA may help to reconcile persistent undertreatment of osteoporosis in NHs, preventing a substantial number of hip fracture and downstream consequences.42,43 Future cost-effectiveness studies focused on NH residents and other frail older adult populations are needed.
Limitations
The findings of our study must be interpreted in light of several limitations. First, because our study was observational, we cannot rule out the possibility of residual confounding. The most plausible mechanisms for this confounding are not immediately clear, though it is possible that the severity of osteoporosis may have guided treatment selection and was unmeasured in our data. Data were unavailable on supplement use (e.g., calcium, vitamin D), the severity of osteoporosis (e.g., T-Score), and laboratory values (e.g., vitamin D levels). However, we did have many important clinical and other covariates measured (e.g., BMI, physical function). We also obtained good balance on 125 measured baseline covariates across treatment groups after IPTW. A second limitation is that we were unable to conduct analyses of subgroups to assess how treatment effects might vary across subgroups defined by important variables like age, cognition, functional status, and life expectancy. Despite the fact that our sample size was larger than some trials of osteoporosis treatments, subgroup analyses were not sufficiently powered due to the incidence of hip fracture events. Third, we were unable to conduct analyses examining safety events (e.g., atypical fractures, hypocalcemia, osteonecrosis of the jaw) or non-hospitalized hip fractures due to the difficulty of ascertaining these events accurately in secondary data.
Conclusions
Denosumab and ZA may be as effective as teriparatide for hip fracture prevention in frail older adults. Given their lower cost and easier administration, denosumab and ZA are likely preferable non-oral treatments for many frail, older adults, especially those residing long-term in NHs.
Supplementary Material
Acknowledgments:
Funding Sources: This study was supported by grants R01AG045441, RF1AG061221, R01AG065722, and R21AG061632 from the National Institute on Aging (NIA) and by grant U54GM1156775 from the National Institute of General Medical Sciences (NIGMS), which funds Dr. Zullo and Advance Clinical and Translational Research (Advance-CTR). Dr. Zullo is also supported by a Department of Veterans Affairs Office of Academic Affiliations Advanced Fellowship in Health Services Research and Development. Dr. Lary is supported, in part, by grant P20GM121301 from NIGMS (PI Lucy Liaw). Role of the Funding Sources: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of a an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.
Conflicts of Interest: Dr. Berry previously received grant money from Amgen unrelated to the current project. Dr. Kiel previously received grant support from Merck Sharp & Dohme and Amgen for clinical trials of osteoporosis treatments and grant funding for investigator initiated research from Amgen and Radius Health unrelated to the current project. All other authors have no relevant conflicts of interest to report.
REFERENCES
- 1.Berry SD, Lee Y, Zullo AR, Kiel DP, Dosa D, Mor V. Incidence of Hip Fracture in U.S. Nursing Homes. The journals of gerontology Series A, Biological sciences and medical sciences. 2016;71(9):1230–1234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Berry SD, Daiello LA, Lee Y, et al. Secular Trends in the Incidence of Hip Fracture Among Nursing Home Residents. J Bone Miner Res. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chandler JM, Zimmerman SI, Girman CJ, et al. Low bone mineral density and risk of fracture in white female nursing home residents. JAMA : the journal of the American Medical Association. 2000;284(8):972–977. [DOI] [PubMed] [Google Scholar]
- 4.Zullo AR, Zhang T, Banerjee G, et al. Facility and State Variation in Hip Fracture in U.S. Nursing Home Residents. Journal of the American Geriatrics Society. 2018;66(3):539–545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Banerjee G, Zullo AR, Berry SD, et al. Geographic Variation in Hip Fracture Among United States Long-Stay Nursing Home Residents. Journal of the American Medical Directors Association. 2016;17(9):865 e861–863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zullo AR, Zhang T, Lee Y, et al. Effect of Bisphosphonates on Fracture Outcomes Among Frail Older Adults. Journal of the American Geriatrics Society. 2019;67(4):768–776. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cummings SR, San Martin J, McClung MR, et al. Denosumab for prevention of fractures in postmenopausal women with osteoporosis. N Engl J Med. 2009;361(8):756–765. [DOI] [PubMed] [Google Scholar]
- 8.Reid IR, Horne AM, Mihov B, et al. Fracture Prevention with Zoledronate in Older Women with Osteopenia. N Engl J Med. 2018;379(25):2407–2416. [DOI] [PubMed] [Google Scholar]
- 9.Neer RM, Arnaud CD, Zanchetta JR, et al. Effect of parathyroid hormone (1–34) on fractures and bone mineral density in postmenopausal women with osteoporosis. The New England journal of medicine. 2001;344(19):1434–1441. [DOI] [PubMed] [Google Scholar]
- 10.Boonen S, Adachi JD, Man Z, et al. Treatment with denosumab reduces the incidence of new vertebral and hip fractures in postmenopausal women at high risk. The Journal of clinical endocrinology and metabolism. 2011;96(6):1727–1736. [DOI] [PubMed] [Google Scholar]
- 11.Lyles KW, Colon-Emeric CS, Magaziner JS, et al. Zoledronic acid and clinical fractures and mortality after hip fracture. The New England journal of medicine. 2007;357(18):1799–1809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Durden E, Pinto L, Lopez-Gonzalez L, Juneau P, Barron R. Two-year persistence and compliance with osteoporosis therapies among postmenopausal women in a commercially insured population in the United States. Arch Osteoporos. 2017;12(1):22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Greenspan SL, Perera S, Ferchak MA, Nace DA, Resnick NM. Efficacy and safety of single-dose zoledronic acid for osteoporosis in frail elderly women: a randomized clinical trial. JAMA internal medicine. 2015;175(6):913–921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kendler DL, Marin F, Zerbini CAF, et al. Effects of teriparatide and risedronate on new fractures in post-menopausal women with severe osteoporosis (VERO): a multicentre, double-blind, double-dummy, randomised controlled trial. Lancet. 2018;391(10117):230–240. [DOI] [PubMed] [Google Scholar]
- 15.Tsai JN, Uihlein AV, Lee H, et al. Teriparatide and denosumab, alone or combined, in women with postmenopausal osteoporosis: the DATA study randomised trial. Lancet. 2013;382(9886):50–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Tan W, Sun J, Zhou L, Li Y, Wu X. Randomized trial comparing efficacies of zoledronate and alendronate for improving bone mineral density and inhibiting bone remodelling in women with post-menopausal osteoporosis. J Clin Pharm Ther. 2016;41(5):519–523. [DOI] [PubMed] [Google Scholar]
- 17.Kendler DL, Roux C, Benhamou CL, et al. Effects of denosumab on bone mineral density and bone turnover in postmenopausal women transitioning from alendronate therapy. J Bone Miner Res. 2010;25(1):72–81. [DOI] [PubMed] [Google Scholar]
- 18.Zebaze RM, Libanati C, Austin M, et al. Differing effects of denosumab and alendronate on cortical and trabecular bone. Bone. 2014;59:173–179. [DOI] [PubMed] [Google Scholar]
- 19.Saliba D, Jones M, Streim J, Ouslander J, Berlowitz D, Buchanan J. Overview of significant changes in the Minimum Data Set for nursing homes version 3.0. Journal of the American Medical Directors Association. 2012;13(7):595–601. [DOI] [PubMed] [Google Scholar]
- 20.Intrator O, Hiris J, Berg K, Miller SC, Mor V. The residential history file: studying nursing home residents’ long-term care histories(*). Health services research. 2011;46(1 Pt 1):120–137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Huitfeldt A, Hernan MA, Kalager M, Robins JM. Comparative Effectiveness Research Using Observational Data: Active Comparators to Emulate Target Trials with Inactive Comparators. EGEMS (Wash DC). 2016;4(1):1234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hernan MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American journal of epidemiology. 2016;183(8):758–764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lage DE, DuMontier C, Lee Y, et al. Potentially burdensome end-of-life transitions among nursing home residents with poor-prognosis cancer. Cancer. 2020;126(6):1322–1329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Noone AM, Lund JL, Mariotto A, et al. Comparison of SEER Treatment Data With Medicare Claims. Medical care. 2016;54(9):e55–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Curtis JR, Xie F, Yun H, Saag KG, Chen L, Delzell E. Risk of hospitalized infection among rheumatoid arthritis patients concurrently treated with a biologic agent and denosumab. Arthritis Rheumatol. 2015;67(6):1456–1464. [DOI] [PubMed] [Google Scholar]
- 26.Curtis JR, Xie F, Chen R, et al. Identifying newly approved medications in Medicare claims data: a case study using tocilizumab. Pharmacoepidemiology and drug safety. 2013;22(11):1214–1221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Berry SD, Zullo AR, McConeghy K, Lee Y, Daiello L, Kiel DP. Defining hip fracture with claims data: outpatient and provider claims matter. Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 2017;28(7):2233–2237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Ray WA, Griffin MR, Fought RL, Adams ML. Identification of fractures from computerized Medicare files. Journal of clinical epidemiology. 1992;45(7):703–714. [DOI] [PubMed] [Google Scholar]
- 29.Zullo AR, Sorial MN, Lee Y, Lary CW, Kiel DP, Berry SD. Predictors of Hip Fracture Despite Treatment with Bisphosphonates among Frail Older Adults. Journal of the American Geriatrics Society. 2020;68(2):256–260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Berry SD, Zullo AR, Lee Y, et al. Fracture Risk Assessment in Long-term Care (FRAiL): Development and Validation of a Prediction Model. The journals of gerontology Series A, Biological sciences and medical sciences. 2018;73(6):763–769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Berry SD, Dufour AB, Travison TG, et al. Changes in bone mineral density (BMD): a longitudinal study of osteoporosis patients in the real-world setting. Arch Osteoporos. 2018;13(1):124. [DOI] [PubMed] [Google Scholar]
- 32.By the American Geriatrics Society Beers Criteria Update Expert P. American Geriatrics Society 2015 Updated Beers Criteria for Potentially Inappropriate Medication Use in Older Adults. Journal of the American Geriatrics Society. 2015. [DOI] [PubMed] [Google Scholar]
- 33.Tsai JN, Uihlein AV, Burnett-Bowie SA, et al. Comparative effects of teriparatide, denosumab, and combination therapy on peripheral compartmental bone density, microarchitecture, and estimated strength: the DATA-HRpQCT Study. J Bone Miner Res. 2015;30(1):39–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lyu H, Zhao SS, Yoshida K, et al. Comparison of Teriparatide and Denosumab in Patients Switching From Long-Term Bisphosphonate Use. The Journal of clinical endocrinology and metabolism. 2019;104(11):5611–5620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Barrionuevo P, Kapoor E, Asi N, et al. Efficacy of Pharmacological Therapies for the Prevention of Fractures in Postmenopausal Women: A Network Meta-Analysis. The Journal of clinical endocrinology and metabolism. 2019;104(5):1623–1630. [DOI] [PubMed] [Google Scholar]
- 36.Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. The journals of gerontology Series A, Biological sciences and medical sciences. 2001;56(3):M146–156. [DOI] [PubMed] [Google Scholar]
- 37.Fried LP, Ferrucci L, Darer J, Williamson JD, Anderson G. Untangling the concepts of disability, frailty, and comorbidity: implications for improved targeting and care. The journals of gerontology Series A, Biological sciences and medical sciences. 2004;59(3):255–263. [DOI] [PubMed] [Google Scholar]
- 38.Rockwood K, Mitnitski A. Frailty defined by deficit accumulation and geriatric medicine defined by frailty. Clin Geriatr Med. 2011;27(1):17–26. [DOI] [PubMed] [Google Scholar]
- 39.Black DM, Greenspan SL, Ensrud KE, et al. The effects of parathyroid hormone and alendronate alone or in combination in postmenopausal osteoporosis. The New England journal of medicine. 2003;349(13):1207–1215. [DOI] [PubMed] [Google Scholar]
- 40.Diez-Perez A, Marin F, Eriksen EF, Kendler DL, Krege JH, Delgado-Rodriguez M. Effects of teriparatide on hip and upper limb fractures in patients with osteoporosis: A systematic review and meta-analysis. Bone. 2019;120:1–8. [DOI] [PubMed] [Google Scholar]
- 41.Tu KN, Lie JD, Wan CKV, et al. Osteoporosis: A Review of Treatment Options. P T. 2018;43(2):92–104. [PMC free article] [PubMed] [Google Scholar]
- 42.Zarowitz BJ, Cheng LI, Allen C, O’Shea T, Stolshek B. Osteoporosis prevalence and characteristics of treated and untreated nursing home residents with osteoporosis. Journal of the American Medical Directors Association. 2015;16(4):341–348. [DOI] [PubMed] [Google Scholar]
- 43.Lind KE, Jorgensen ML, Gray LC, Georgiou A, Westbrook JI. Anti-osteoporosis Medication Use in a High Fracture-Risk Population: Contemporary Trends in Australian Residential Aged Care Facilities. Health Serv Insights. 2019;12:1178632919852111. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.