Comparisons Between Different Anti-osteoporosis Medications on Postfracture Mortality: A Population-Based Study

Abstract

Context

Osteoporosis is becoming a global epidemic in aging societies. Anti-osteoporotic medications can prevent fractures, and their pleiotropic effect on mortality is interesting but not well compared among each other.

Objective

To provide real-world evidence on the pleiotropic effect of different anti-osteoporotic medications on all-cause mortality, stratified by fracture site, sex, and age.

Methods

This longitudinal population-based postfracture cohort study, included mega-data from subjects ≥40 years of age with osteoporotic fracture who used anti-osteoporotic medications as recorded in Taiwan's National Health Insurance Research Database from 2009 to 2017 and followed until 2018. A multivariate Cox proportional hazards model with immortal time bias was used to assess the relationship between fracture sites and mortality stratified by anti-osteoporosis medication.

Results

A total of 46 729 subjects with an average age of 74.45 years (80.0% female) and a mean follow-up period of 4.73 years were enrolled. In the total fracture group, compared with raloxifene and bazedoxifene, we found that alendronate/risedronate (hazard ratio [HR] 0.83; 95% CI, 0.79-0.88), denosumab (HR 0.86; 95% CI, 0.81-0.91), and zoledronic acid (HR 0.78; 95% CI, 0.73-0.84) resulted in significantly lower mortality. Similar trends were observed in the hip, vertebral, or nonhip/nonvertebral fracture groups. Subjects receiving long-acting zoledronic acid showed the lowest mortality in the subanalysis according to sex or age over 65 years.

Conclusion

This real-world mega-data study suggests that the usage of osteoporotic medication, especially a long-acting regimen, may lower postfracture mortality.

osteoporosis, human association studies, anti-osteoporosis medication, therapeutics, mortality

Osteoporosis is the most common bone-related disease and a major public health concern (1). With the rapidly aging population worldwide, the incidence and prevalence of osteoporosis have significantly increased (2). There have been remarkable achievements in the diagnosis and treatment of osteoporosis in recent years. Improving the knowledge of the mechanisms of age-related bone loss drove the development of anti-osteoporosis medications (AOMs) from the 1960s (3). To date, several studies have reported many bone-related effects, such as increasing bone mineral density, decreasing bone loss, and prevention of osteoporotic fractures (4). Thus, many guidelines recommend that osteoporosis treatment should be considered for those who have had fracture or who have a high risk of fracture (5, 6).

Approved AOMs include antiresorptive and bone-forming drugs. Only antiresorptive agents can be applied for long-term use. A recent systematic review (7-9) and meta-analyses found that AOMs have some nonbone effects, such as reducing all-cause and cardiovascular mortality, myeloma, and diabetes mellitus (7, 10-12). However, except for randomized clinical trials (RCTs), only limited observational studies have demonstrated these efficacies, and only a small number have conducted comparisons between a few types of AOMs (13-16). One meta-analysis study even found that bisphosphonate treatment could not reduce overall mortality (8).

There are concerns regarding these studies, such as inadequate sample size, younger study populations, limited studies conducted with male subjects, and narrow inclusion criteria (eg, types of fracture, types of AOMs, and duration of treatment), which has led to unsatisfactory RCTs and clinical trials. Furthermore, the different nonbone effects of AOMs may also be clarified in well-controlled observational cohort studies. Therefore, more observational studies from the real world are needed to make up for the unmet need from traditional clinical trial studies, such as population-based databases, head-to-head comparisons of different types of AOMs, and longitudinal follow-up studies (17).

In this study, the primary objective was to compare the real-world evidence of pleiotropic effects on long-term mortality between different AOMs by using Taiwan's National Health Insurance Research Database. The secondary objective was to recognize the nonbone effects of AOMs in subgroups according to sex and age greater than 65 years.

Methods

Data Sources

This longitudinal population-based cohort study was conducted using the National Health Insurance Research Database (NHIRD) provided by the Health and Welfare Data Science Center (HWDC) of the Ministry of Health and Welfare (MOHW), including “Ambulatory Care Expenditures by Visits” and “Inpatient Expenditures by Admissions,” interlinking “Cause of Death Data” from the national death registry to identify death status.

Taiwan National Health Insurance (NHI) was launched in 1995 and was implemented until 2018, and the coverage rate was 99.99% among the population of 23 million people; therefore, NHIRD could provide complete health care medical records and real-world evidence with full representation of the Taiwanese population (18). Disease diagnosis was performed according to the International Classification of Diseases, 9th and 10th Revision, Clinical Modification (ICD-9/10-CM) diagnostic codes, which were replaced with ICD-10-CM in 2016 for reporting diagnosis on medical claims.

Mortality status was identified from the national death registry system. Data included are a single underlying cause of death (ICD-10-CM) and demographic data from death certificates.

Cohort Establishment

We identified all patients aged 40 years and older who were first diagnosed with osteoporosis (ICD-9/10-CM: 733.0-733.1/M80-M81) based on ambulatory claims and total fracture (ICD-9/10-CM: 800-829, 733.1/S02-S92) from hospital claims, as described and well validated in detail in previous research, and who were prescribed at least one AOM agent based on ambulatory claims between January 2009 and December 2017 (19, 20). We applied backward washout through 2008 to confirm that there was no osteoporosis or fracture diagnosis or AOM history before 2009. The index date was defined as the date of the first hospitalization with total fracture. Finally, we excluded those who died prior to the entry date. To confirm the effect of AOMs, we conducted a sensitivity analysis of various fracture sites, including hip, vertebral, and nonhip and nonvertebral fractures. Those diagnoses of fracture were claimed in the hospital records during the follow-up period.

Exposure to Anti-osteoporosis Medications

In this study, the use of AOMs was defined as the use of osteoporosis medications approved by the Taiwan Food and Drug Administration (TFDA), including bisphosphonates (weekly oral: alendronate and risedronate; quarterly intravenous: ibandronate; and annual intravenous: zoledronic acid), selective estrogen-receptor modulators (SERMs; daily oral: raloxifene and bazedoxifene), and RANKL inhibitors (biannual injection: denosumab) and excluding those regimens used for the purpose of malignant neoplasm or Paget's disease. According to the limited duration of usage in NHI reimbursement regulations, we excluded patients treated with the anabolic agent (teriparatide) or salmon synthetic (calcitonin). Patients treated with SERMs (raloxifene and bazedoxifene) were used as the reference group due to the neutral effect of SERMs on all-cause mortality; these comparisons were conducted to assess the associations of longer-acting and short-acting agents with mortality (21).

To identify the single types of AOMs, we checked all prescriptions of all outpatient visits in the follow-up period and excluded patients who had switched between any type of AOMs. To account for the potential immortal time bias, the day of initiating AOMs, rather than the day of fracture diagnosis, was used to calculate the follow-up interval for estimating the survival effect conservatively.

Covariables

Baseline characteristics considered in the study, including sex and age but not lifestyle factors, were collated on the first date of fracture diagnosis. Mortality after an osteoporotic fracture may be attributable to a patient's known health factors; therefore, we calculated the Charlson Comorbidity Index (CCI) score to adjust the impact of illness before fracture. The CCI score was calculated using the ambulatory visit diagnoses of CCI diseases with the ICD-9/10-CM codes within 1 year before the index date (22).

Statistics

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.

The analyses of baseline characteristics were performed using t tests for continuous variables and χ² tests for categorical variables. The multivariable Cox proportional hazard model was used to estimate the hazard ratio (HR) with a 95% CI of significant differences between the single type of AOMs according to several categorical variables. To determine the role of the type of AOMs on mortality risk, raloxifene/bazedoxifene, and zoledronic acid were separately compared to other types of AOMs using a Cox proportional hazards model adjusted for any variable that became unbalanced after stratification. We conducted a subgroup analysis stratified by sex and age greater than 65 years to determine the efficacy of AOMs among those subjects.

A two-sided probability value of 0.05 was used to indicate statistical significance. All data were analyzed with SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Between 2009 and 2017, we found 219 461 patients with new diagnoses of osteoporotic fractures in the NHIRD, which prospectively collected medical care data from the universal coverage population. Patients were excluded if they were younger than 40 years old on the index date, ever used osteoporosis medication before the index date, ever switched the type of AOMs, or used calcitonin and teriparatide.

A total of 46 729 osteoporotic fracture patients aged 40 years and older in Taiwan were included between 2009 and 2017, as summarized in Table 1. The mean age of the subjects was 74.45 years, patients were predominantly female (80.0%), and 32% died during the follow-up period. The most frequent type of AOMs in both surviving and nonsurviving subjects was alendronate/risedronate, followed by denosumab and raloxifene/bazedoxifene.

Table 1.

Characteristics of study participants at baseline

	Surviving group (n = 31 707)	Nonsurviving group (n = 15 022)	P
Sex			<0.0001
ȃMale	5042 (15.9%)	4311 (28.7%)
ȃFemale	26 665 (84.1%)	10 711 (71.3%)
Age at baseline (years)	72.50 ± 9.22	78.56 ± 7.05	<0.0001
CCI	1.77 ± 1.86	2.44 ± 2.15	<0.0001
Follow-up period (years)	5.23 ± 2.48	3.70 ± 2.31	<0.0001
Duration of osteoporosis medication (years)	1.05 ± 1.05	0.76 ± 0.84	<0.0001
Interval month from fracture to osteoporosis treatment (months)	22.42 ± 0.13	15.34 ± 0.13	<0.0001
Type of osteoporosis medications			<0.0001
ȃRaloxifene/bazedoxifene	4489 (14.2%)	2138 (14.2%)
ȃAlendronate/risedronate	12 052 (38.0%)	8410 (56.0%)
ȃIbandronate	1754 (5.5%)	607 (4.0%)
ȃZoledronic acid	3495 (11.0%)	1222 (8.1%)
ȃDenosumab	9917 (31.3%)	2645 (17.6%)
Total fracturea	31 707 (67.9%)	15 022 (32.2%)
Hip fracture	10 925 (60.3%)	7192 (39.7%)
Vertebral fracture	11 858 (68.2%)	5540 (31.8%)
Nonhip/nonvertebral fracture	9050 (79.1%)	2390 (20.9%)

Data are n (%) for categorical variables or mean ± SD for continuous variables, unless indicated otherwise. P value for t test on continuous variables, or χ² test on categorical variables.

Abbreviation: CCI, Charlson Comorbidity Index.

Total fracture includes hip, vertebral, forearm, and shoulder fractures. Subjects might have multiple types of fractures in a single hospital claim.

After adjusting for immortal time bias, with the exception of ibandronate, nearly all AOMs had a lower mortality risk than SERMs (as a reference) in all 4 types of fracture sites (Table 2). At all fracture sites, compared with SERMs, alendronate/risedronate (HR 0.83, 95% CI 0.79-0.88), denosumab (HR 0.86, 95% CI 0.81-0.91), and zoledronic acid (HR 078, 95% CI 0.73-0.84) significantly lowered mortality risk, and the same finding was observed in the hip and vertebral fracture groups. Females had a lower mortality risk than males, and the risk of mortality increased with age.

Table 2.

Multivariate Cox proportional hazard analyses of the association between fracture sites and all-cause mortality

	Total fracture	Hip fracture	Vertebral fracture	Nonhip/nonvertebral fracture
Gender (ref. male)	1.00	1.00	1.00	1.00
ȃFemale	0.61 (0.59-0.63)**	0.65 (0.61-0.68)**	0.61 (0.57-0.64)**	0.52 (0.47-0.58)**
Age	1.08 (1.08-1.08)**	1.07 (1.07-1.08)**	1.08 (1.07-1.08)**	1.08 (1.07-1.09)**
CCI	1.13 (1.13-1.14)**	1.12 (1.10-1.13)**	1.14 (1.12-1.15)**	1.18 (1.16-1.20)**
Type of osteoporosis medications
ȃRaloxifene/bazedoxifene (ref.)	1.00	1.00	1.00	1.00
ȃAlendronate/risedronate	0.83 (0.79-0.88)**	0.78 (0.72-0.83)**	0.88 (0.81-0.95)**	0.88 (0.79-0.99)**
ȃIbandronate	0.96 (0.88-1.05)	0.91 (0.79-1.04)	0.99 (0.85-1.14)	1.05 (0.85-1.29)
ȃDenosumab	0.86 (0.81-0.91)**	0.79 (0.73-0.86)**	0.89 (0.81-0.98)**	0.96 (0.84-1.10)
ȃZoledronic acid	0.78 (0.73-0.84)**	0.78 (0.70-0.86)**	0.76 (0.67-0.86)**	0.81 (0.68-0.97)**

Data presented as adjusted hazard ratios (95% CI), adjusted with immortal time bias.

Abbreviation: CCI, Charlson Comorbidity Index.

**P < 0.05.

As shown in Table 3, the Cox proportional hazards regression models revealed the benefits of AOMs in females. In the total fracture group of females, compared with SERMs, alendronate/risedronate (HR 0.84; 95% CI, 0.80-0.88), denosumab (HR 0.84; 95% CI, 0.79-0.89), and zoledronic acid (HR 0.78; 95% CI, 0.72-0.85) were associated with a significantly lower risk of mortality. In contrast, using zoledronic acid as the reference, ibandronate was associated with the highest mortality risk, followed by either alendronate or denosumab in different types of fracture sites (Table 4).

Table 3.

Multivariate Cox proportional hazard analyses of the association between fracture sites and all-cause mortality in females

	Total fracture	Hip fracture	Vertebral fracture	Nonhip/nonvertebral fracture
Age	1.09 (1.08-1.09)**	1.08 (1.08-1.09)**	1.08 (1.08-1.09)**	1.09 (1.09-1.10)**
CCI	1.15 (1.14-1.16)**	1.13 (1.12-1.15)**	1.16 (1.14-1.17)**	1.19 (1.17-1.22)**
Type of osteoporosis medications
ȃRaloxifene/bazedoxifene (ref.)	1.00	1.00	1.00	1.00
ȃAlendronate/risedronate	0.84 (0.80-0.88)**	0.78 (0.73-0.85)**	0.88 (0.81-0.96)**	0.89 (0.79-1.00)**
ȃIbandronate	0.96 (0.87-1.05)	0.91 (0.79-1.05)	0.98 (0.84-1.14)	1.05 (0.84-1.31)
ȃDenosumab	0.84 (0.79-0.89)**	0.78 (0.71-0.85)**	0.88 (0.80-0.98)**	0.91 (0.79-1.05)
ȃZoledronic acid	0.78 (0.72-0.85)**	0.77 (0.69-0.86)**	0.78 (0.68-0.89)**	0.80 (0.65-0.98)**

Data presented as adjusted hazard ratios (95% CI), adjusted for immortal time bias.

Abbreviation: CCI = Charlson Comorbidity Index.

**P < 0.05.

Table 4.

Multivariate Cox proportional hazard analyses of the association between fracture sites and all-cause mortality in females

	Total fracture	Hip fracture	Vertebral fracture	Nonhip/nonvertebral fracture
Age	1.09 (1.09-1.09)**	1.08 (1.08-1.09)**	1.08 (1.08-1.09)**	1.09 (1.08-1.10)**
CCI	1.15 (1.14-1.16)**	1.13 (1.11-1.14)**	1.15 (1.14-1.17)**	1.19 (1.16-1.22)**
Type of osteoporosis medications
ȃZoledronic acid (reference)	1.00	1.00	1.00	1.00
ȃAlendronate/risedronate	1.07 (1.00-1.15)	1.01 (0.91-1.12)	1.13 (1.01-1.28)**	1.11 (0.92-1.34)
ȃIbandronate	1.23 (1.10-1.37)**	1.19 (1.02-1.40)**	1.26 (1.06-1.50)**	1.31 (1.01-1.71)**
ȃDenosumab	1.08 (1.00-1.17)	1.02 (0.91-1.14)	1.14 (1.00-1.31)	1.14 (0.93-1.40)

Data presented as adjusted hazard ratios (95% CI), adjusted for immortal time bias.

Abbreviation: CCI, Charlson Comorbidity Index.

**P < 0.05.

As SERMs are not recommended in males with osteoporosis, zoledronic acid was used as the reference for further analysis in males. Table 5 shows the higher mortality risk associated with denosumab in total fracture (HR 1.17; 95% CI, 1.03-1.33), vertebral fracture (HR 1.28; 95% CI, 1.01-1.61) sites, and ibandronate in vertebral fracture (HR 1.54; 95% CI, 1.04-2.30) sites.

Table 5.

Multivariate Cox proportional hazard analyses of the association between fracture sites and all-cause mortality in males

	Total fracture	Hip fracture	Vertebral fracture	Nonhip/nonvertebral fracture
Age	1.06 (1.05-1.06)**	1.06 (1.05-1.07)**	1.06 (1.05-1.06)**	1.05 (1.04-1.06)**
CCI	1.09 (1.08-1.11)**	1.08 (1.06-1.11)**	1.09 (1.07-1.12)**	1.13 (1.09-1.18)**
Type of osteoporosis medications
ȃZoledronic acid (ref.)	1.00	1.00	1.00	1.00
ȃAlendronate/risedronate	1.05 (0.94-1.18)	0.97 (0.83-1.13)	1.20 (0.99-1.46)	0.99 (0.73-1.34)
ȃIbandronate	1.23 (0.94-1.61)	1.10 (0.70-1.73)	1.54 (1.04-2.30)**	1.03 (0.54-1.99)
ȃDenosumab	1.17 (1.03-1.33)**	1.06 (0.89-1.26)	1.28 (1.01-1.61)**	1.29 (0.92-1.81)

Data presented as adjusted hazard ratios (95% CI), adjusted for immortal time bias.

Abbreviation: CCI, Charlson Comorbidity Index.

**P < 0.05.

In the group older than than 65 years, the total fracture site patients treated with either alendronate/risedronate (HR 0.82; 95% CI, 0.78-0.86), denosumab (HR 0.86; 95% CI, 0.81-0.91), or zoledronic acid (HR 0.79; 95% C, 0.73-0.85) had a significantly lower mortality risk, and the hip, vertebral, nonhip/nonvertebral fracture sites also exhibited similar results (Table 6).

Table 6.

Multivariate Cox proportional hazard analyses of the association between fracture sites and all-cause mortality in age ≥ 65 years group

	Total fracture	Hip fracture	Vertebral fracture	Nonhip/nonvertebral fracture
Gender (ref. male)	1.00	1.00	1.00	1.00
ȃFemale	0.58 (0.56-0.60)**	0.65 (0.61-0.68)**	0.57 (0.53-0.60)**	0.50 (0.45-0.56)**
CCI	1.12 (1.11-1.12)**	1.10 (1.08-1.11)**	1.12 (1.10-1.13)**	1.17 (1.15-1.19)**
Type of osteoporosis medications
ȃRaloxifene/bazedoxifene (ref.)	1.00	1.00	1.00	1.00
ȃAlendronate/risedronate	0.82 (0.78-0.86)**	0.76 (0.71-0.82)**	0.85 (0.78-0.93)**	0.87 (0.77-0.98)**
ȃIbandronate	0.98 (0.90-1.08)	0.94 (0.82-1.08)	1.00 (0.86-1.16)	1.10 (0.89-1.37)
ȃDenosumab	0.86 (0.81-0.91)**	0.80 (0.74-0.88)**	0.89 (0.81-0.99)**	0.90 (0.78-1.04)
ȃZoledronic acid	0.79 (0.73-0.85)**	0.79 (0.71-0.87)**	0.76 (0.67-0.86)**	0.80 (0.66-0.96)**

Data presented as adjusted hazard ratios (95% CI), adjusted for immortal time bias.

Abbreviation: CCI = Charlson Comorbidity Index.

**P < 0.05.

Discussion

This is the first nationwide population-based study to compare the risk of all-cause mortality after osteoporotic fracture between nearly all types of AOMs. This real-world mega-study demonstrates the association between the usage of AOMs and reduced all-cause mortality risk for all fracture sites, regardless of sex or age, using the Taiwan NHIRD involving 46 729 patients aged 40 years and older who had suffered osteoporotic fracture, with a mean of 4.73 follow-up years.

Many placebo-controlled RCTs have reported the efficacy of AOMs (8, 10). RCTs are indeed the “gold standard” research design; however, they may have limitations, such as a limited sample size, short-term follow-up time, low external validity, and ethical issues (23). Another limitation of RCTs is that they usually compare the effects between 2 types of AOMs (24, 25). The comparison of all types of AOMs concomitantly is an unmet need that can only be satisfied using real-world evidence from mega-data studies. The US Food and Drug Administration (FDA) recently actively evaluated the real-world evidence used to explain clinical uncertainties and facilitate regulatory decisions (26). From the real-world evidence of this study, the policy of regulatory decisions for osteoporosis treatment can be considered in advance, especially in an aging society.

The mechanism of reduced postfracture all-cause mortality by AOMs has been postulated, but heterogeneous findings have been reported (7, 10). The pleiotropic effects of AOMs on cancer, cardiovascular risk, pneumonia, diabetes, stroke, or even DNA damage may contribute to reduce mortality (27-29). AOMs will lower fracture risk but do not eliminate fractures. Therefore, the influence of refracture on mortality rate is also of concern. In much of the world, a patient who fractures while on an AOM will not necessarily be changed to another medication. However, subjects may consider switching AOMs after refracture in real-world clinical practice. To discriminate the different effect of AOMs on mortality, patients who had ever used more than one kind of anti-osteoporosis medication were excluded in this study. As the refracture rate was less than 5% in our cohort, the contribution of refracture to mortality between AOMs is minimal. Despite the structural distinctions, such as the type, potency, dosage, and administration route of the AOM, mortality decreased to a greater extent along with the longer duration of drug frequency among the different types of AOMs in this study. The adjusted hazard ratios for ibandronate are ranged from 0.91 to 1.10 without statistical significance. The reason for no difference relative to raloxifene/bazedoxifene may be due to the limited case number of ibandronate recipients (∼5% of study subjects). Whether health behavior (patient adherence or preference) might also play an important role deserves further study.

This study had some strengths that should be emphasized. First, because the nationwide databank covers 99.7% of the total Taiwanese population, this study overcomes the limitations of previous observational or cohort studies and provides reliable comprehensive real-world evidence. Second, the average follow-up period of 4.73 years was longer than those in the latest RCTs or observational studies. Kanis JA et al reported a high mortality risk in the year following fractures, but the mortality risk was stabilized after more than 1 year of follow-up in this study (30). Third, we assessed the pleiotropic effects by comparing all approved antiresorptive effects at a time. Although some dosages of AOMs are not well studied, most AOMs exhibit favorable effects in the case of long-term usage. Fourth, this study examined the sex- and age-specific effects of AOMs and showed consistent findings. Finally, we compared the pleiotropic effects of AOMs on mortality at different sites of fracture. Therefore, this study could support clinical treatment decisions, regardless of the AOM type or fracture site, and help to close the treatment gap. These differences can be considered when selecting AOMs in an individualized scenario. However, ongoing assessments of new AOMs may provide a greater number of options in the future.

Owing to the data limitation of the NHIRD, there were unmeasured potential confounders, such as health inequality or health literacy. Body mass index (BMI), vitamin D, smoking, and alcohol are known risk factors for osteoporosis and some chronic diseases; therefore, we calculated the CCI among chronic diseases to replace these factors not available directly in the NHIRD. Based on the NHI regulations, those AOMs have the same indications and payment regulations. This could decrease the confounder bias of disease severity. We performed a sensitivity analysis using the E-value to assess the unmeasured confounding variables, and the result was 1.20 to 1.66, indicating the study's stability and reliability with regard to the unmeasured variables (31). However, the impact of lifestyle and vitamin D level should be of concern always. Moreover, risedronate has been on the Taiwan market since 2016, and there were too few subjects to be categorized into a single group. Given the same route of administration (ie, oral) and frequency of the treatment (ie, weekly), risedronate was classified with alendronate into one group in this study. As no oral ibandronate was available during the study period, future research is needed.

In conclusion, treatment for osteoporosis has the potential to minimize mortality risk in people of all ages and sexes for any type of fracture. The longer-acting treatments could lower mortality risk. To make better osteoporosis therapeutic decisions, additional real-world evidence studies on treatment efficacy are required to close the treatment gap.

Acknowledgments

We are grateful to the Health Data Science Center, National Cheng Kung University Hospital for providing administrative and technical support.

Funding

This study was funded by research grants MOST 106-2314-B-006-064-MY2, MOST 108-2314-B-006-043-MY2, and MOST 110-2314-B-006-054-MY2 from the Ministry of Science and Technology, Taiwan, partially supported by a research grant TOA-2020-B-01 from the Taiwanese Osteoporosis Association and grant NCKUH-10909042, NCKUH-11103023, and NCKUH-11103054 from National Cheng Kung University Hospital, Taiwan. The study (ClinicalTrials.gov Identifier: NCT05366621) was approved by the Institutional Review Board (IRB) of the National Cheng Kung University Hospital (NCKUH) (Nos. A-ER-106-123, B-ER-109-346, and B-ER-110-124).

Author Contributions

Conception and design of the study: C.H.W., C.C.L., Y.F.C., and J.S.H. Statistical analyses: C.C.L. and F.W.L. Research data interpretation: all authors. Acquisition of data: C.H.W., C.C.L., and Y.H.H. Suggestions and discussion: all authors. Drafting the article: C.C.L. and C.H.W. Critical revision of the manuscript: all authors. Final approval of the manuscript: all authors.

Disclosures

L.C.C., H.Y.H., and L.F.W. have nothing to declare. C.Y.F. has honoraria for lectures from Amgen, Merck, Alvogen/Lotus. H.J.S. has honoraria for lectures from Eli Lilly, Roche, Amgen, Merck, Alvogen/Lotus. W.C.H. has honoraria for lectures, attending meetings, and/or travel from Eli Lilly, Roche, Amgen, Merck, Servier laboratories, GE Lunar, Harvester, TCM Biotech, and Alvogen/Lotus.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories from the Health and Welfare Data Science Center approval by the Department of Statistics, Ministry of Health and Welfare, Taiwan. For ethical and privacy reasons, the data only were accessed and statistically analyzed on Health and Welfare Data Science Center. All datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Abbreviations

 
• AOM

  anti-osteoporosis medication

 
• CCI

  Charlson Comorbidity Index

 
• HR

  hazard ratio

 
• ICD-9/10-CM

  International Classification of Diseases, 9^th/10th Revision, Clinical Modification

 
• NHIRD

  National Health Insurance Research Database (Taiwan)

 
• RCT

  randomized controlled trial

 
• SERMs

  selective estrogen-receptor modulators