A real-world study of treatment patterns among patients with osteoporotic fracture: analysis of a Japanese hospital database

Summary Health records of patients hospitalized for osteoporotic fracture were analyzed. Prior to the index hospital admission, most patients were not receiving any antiosteoporotic treatment. During the index hospitalization visit, 25.5% of patients received antiosteoporotic treatment. The most common treatment regimens were active vitamin D3, bisphosphonates, and teriparatide. Purpose To examine the real-world treatment patterns and factors associated with receipt of treatment among Japanese patients with osteoporotic fracture. Methods
 We retrospectively analyzed health records of patients who were hospitalized for osteoporotic fracture between February 2016 and February 2018 in Japan. The type and duration of treatment with antiosteoporotic medications prescribed during hospital stays and after discharge were examined using descriptive statistics. Demographic and clinical factors (e.g., age, previous diagnoses, Charlson Comorbidity Index scores) associated with osteoporotic treatment were explored using multivariable logistic regression. Results A total of 112,275 patient medical records were evaluated, including 56,574 records from patients with hip fracture, 26,681 records from patients with vertebrae fracture, and 29,020 patients with non-vertebral non-hip fractures. Prior to the index hospital admission, most patients (91.7%, n = 102,919) were not receiving any antiosteoporotic treatment. For those receiving treatment, active vitamin D3 (51.1%, n = 4778) and bisphosphonates (47.5%, n = 4441) were the most common. During the index hospitalization visit, 25.5% (n = 28,678) of patients received treatment for their fracture, including active vitamin D3 (n = 17,074), bisphosphonates (n = 10,007), and teriparatide (n = 4561). Upon discharge, 41.5% (n = 46,536) of patients returned to their home and 34.3% (n = 38,542) of patients were transferred to a different hospital or medical care facility. Variables associated with receipt of treatment at follow-up included older age, previous diagnoses of osteoporosis and fracture, and higher Charlson Comorbidity Index scores. Conclusion Despite osteoporotic fracture being a major health concern within older Japanese populations, treatment with antiosteoporotic medication regimens remains generally low. Supplementary Information The online version contains supplementary material available at 10.1007/s11657-022-01201-x.


Introduction
Osteoporosis and osteoporotic fracture are major health concerns for the estimated 15 million people in Japan with osteoporosis [1], with greater lifetime risk for fractures to the spine and hip as individuals increase in age [2,3].Osteoporotic fracture has been associated with an increased mortality risk of between 10 and 20% in the 1st year after the fracture, and among individuals with osteoporotic fracture, functional status has been found to decrease by approximately 60% when compared with prefracture functional status levels [1].Therefore, it is important to consider pharmacological treatments as preventative measures among patients with diagnosed osteoporosis and adequately treat osteoporotic fracture after it has occurred [4].
Clinical guidelines in Japan recommend pharmacological treatment for individuals with a history of osteoporosis and osteoporotic fracture [5].Despite this recommendation, previous studies have observed that only 20% of individuals in Japan with osteoporosis received antiosteoporotic treatment [1].Additional reports have indicated that only 18.7% of patients were prescribed antiosteoporotic treatment within 1 year of follow-up for a hip fracture [6].More recent studies have examined individuals who are at greater risk for fractures, including women who are perimenopausal or postmenopausal and who have an osteoporosis diagnosis and individuals who have comorbid conditions such as cancer.These studies have found that some patients are unwilling to take medication, while other patients might not be actively receiving care for their osteoporosis; furthermore, the risk for osteoporotic fracture can increase in patients who are not given pharmacological treatment for osteoporosis [7][8][9].However, additional studies assessing real-world data are needed to further understand the treatment patterns of antiosteoporotic therapy following hospitalization for an osteoporotic fracture.
To address these knowledge gaps, this study was conducted to describe the recent treatment landscape among patients in Japan with osteoporotic fracture.We retrospectively analyzed health records of Japanese patients who were hospitalized for osteoporotic fracture to examine the type and duration of treatment with antiosteoporotic agents prescribed to patients during hospital stays and after hospital discharge.Other demographic variables and clinical characteristics, including sex, age, fracture location, previous diagnosis of osteoporosis, and comorbid health conditions, also were examined to further understand their association with treatment patterns.

Study design
The study design was a retrospective database analysis of patients diagnosed and hospitalized for osteoporotic fracture.Patient medical records and administrative data were obtained from the Medical Data Vision (MDV) database in Japan.The MDV is a large database composed of a wide range of administrative and claims data for more than 30 million patients across 400 hospitals [10].MDV primarily contains data from hospitals that use a claims-based payment system, referred to as the Diagnosis Procedure Combination (DPC) system, that is administrated by the Japanese Ministry of Health, Labor, and Welfare.As of December 2018, 18% of DPC hospitals in Japan were included in the MDV database [11].The database contains information such as sex, birth year, department of care, dates of medical encounters, payment plans, diagnoses and medical procedures (defined by International Classification of Diseases, 10th Revision [ICD-10] codes), laboratory orders and results, and prescriptions associated with inpatient and outpatient hospital encounters [10].This study was reviewed by the Public Health Research Foundation Review Board and was determined to be exempt from human research according to the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan.

Study population and timeline
All data were collected from the MDV database between February 1, 2015 and February 28, 2019 (Online Resource 1, Figure S-1).The study analytic cohort was comprised of patients who were at least 50 years old and had been admitted to the hospital with a diagnosis of osteoporotic fracture (the index hospitalization date was defined as the date that a patient was initially admitted to the hospital for his/her fracture and extended to the date of discharge).Eligible patients needed to have at least one visit (documented in their medical record) to the hospital occurring at least 12 months before the index hospitalization (12-month period before the date of their index hospitalization termed baseline period), and at least one visit to the hospital occurring at least 12 months after the date of discharge from the index hospitalization (12-month period after the date of discharge from the index hospitalization termed follow-up period).Accordingly, to allow for the baseline and follow-up periods, the study cohort included patients who were admitted to the hospital with a diagnosis of osteoporotic fracture between February 1, 2016 and February 28, 2018 (Online Resource 1, Figure S-2).

Demographic and clinical characteristics
The primary demographic variables that were documented included sex, age, and the skeletal site of the osteoporotic fracture at the index hospitalization date.Patient data were divided into three mutually exclusive cohorts based on the skeletal site of fracture observed during the index hospitalization: hip, vertebrae (spine), and non-vertebral non-hip (NVNH).NVNH fracture included fractures to the rib(s), sacrum, pubis, upper end of the humerus, lower or upper end of the radius or ulna, forearm, femur, or upper end of the tibia but excluded fractures to the skull, phalanges, and toes.Consistent with previously published literature [12][13][14], patients in all fracture cohorts were selected based on diagnoses using ICD-10 codes (Online Resource 1, Table S-1).Patients with multiple fracture sites during the index hospitalization were excluded from the study.Clinical variables documented during the baseline period included the presence of comorbid health conditions as described by the Charlson Comorbidity Index (CCI) [15].The CCI includes 20 categories of comorbidities, as defined by diagnosis codes, with associated weights corresponding to the severity of the comorbid condition of interest [15].Other documented comorbid conditions (e.g., cerebrovascular disease, congestive heart failure, peripheral vascular disease, and myocardial infarction) were identified by ICD-10 codes and/or by laboratory results.
The variables documented at the index hospitalization date included a previous diagnosis of osteoporosis and/or a previous diagnosis of osteoporotic fracture; smoking index as reported in the MDV database (reported as the number of cigarettes smoked per day multiplied by the number of years of smoking); activities of daily living (ADL) scores as reported in the MDV database, including information on eating, transferring, grooming, use of the toilet/going to the toilet, bathing, level ground walking, stairs, and clothing; route (hospitalized from home, transferred from another hospital, or admitted from another healthcare facility) of admission to the hospital; and the type of hospitalization (planned visit, emergency, or unplanned nonemergency).The clinical variables documented during follow-up included the discharge destination (home, another hospital or healthcare facility, or a geriatric care center) from the hospital, health outcome status (healing/curing of fracture, remission, no change, exacerbation, death, or other) upon discharge, and the presence of home care (care within the same hospital, care within a different hospital, no care, or unknown care status) after discharge.

Treatment patterns
Among the available data, the rates and the class of pharmacological treatment that patients received during the baseline period, index hospitalization, and follow-up were documented.Across these three time points, pharmacological treatment data included the number and percentage of patients receiving an antiosteoporotic agent, as well as the average number of treatment days.Treatment data during the baseline period included the use of antiosteoporotic agents, the class of antiosteoporotic agents received, and the average number of days from the time of treatment initiation to the index hospitalization date.Treatment data during the index hospitalization also included the classes of antiosteoporotic agents received, the number of days from the index hospital admission date to the date on which patients received the first observed antiosteoporotic agent, and the average number of days patients received medications while in the hospital.Finally, treatment data during the follow-up period included the classes of antiosteoporotic agents received, time from hospital discharge to posthospital treatment initiation, the first observed antiosteoporotic agent treatment regimen after discharge (treatment must be received within 14 days after discharge to be included), and the number of patients who either discontinued, switched/restarted, or augmented their treatment regimen.
Treatment discontinuation was defined as the point at which patients had a minimum refill gap (approximately 60 days) between their most recent observed prescription and the end of the 12-month follow-up period.Switching of treatment was defined as the observation of an alternative medication without a refill of the original medication and was based on the observation of prescription fills for other medication classes following the discontinuation date of the original medication.Augmentation was defined as the initiation of a new medication class with continued use of the original medication for a minimum of 60 days of combined use.Adherence to the treatment regimen was also measured using the medication possession ratio (MPR).The MPR was calculated as follows: the sum of the days' supply of medication in an observation period divided by the number of days in the observation period.Categories of MPR (e.g., deciles) were also documented and a dichotomous indicator of adherence was computed.Based on the MPR threshold of 0.8 commonly used to define medication adherence in the osteoporosis literature [16][17][18], a value greater than or equal to 0.80 was considered adherent and a value less than 0.80 was considered nonadherent.Adherence to the first observed treatment regimen was measured by patients' cumulative exposure to medication during the follow-up period.

Statistical analyses
All statistical analyses were conducted using SAS Version 9.4 (SAS Institute; Cary, North Carolina).A descriptive analysis of continuous study variables generated mean values, medians, and standard deviations (SDs) from all available data.Categorical study variables were analyzed with frequency distributions and reported with numbers and percentages.Multivariable logistic regression models were used to examine the associations between patient demographic and clinical characteristics and receipt of treatment, and the outcomes of the associations were presented as odds ratios with 95% confidence intervals (CIs).Independent variables, which included age, sex, CCI score, previous osteoporosis diagnosis, previous osteoporotic fracture diagnosis, discharge destination, and medication use at discharge, were included as covariates in the regression models.

Demographic and clinical characteristics
A total of 112,275 patient medical records were evaluated, including 56,574 records from patients with hip fracture, 26,681 records from patients with vertebrae fracture, and 29,020 of patients with NVNH fractures.The average age of the total patient sample was 80.0 years (SD, 10.5) (Table 1 At the index hospitalization visit, most patients had not been previously diagnosed with osteoporosis (94.4%, n = 105,976) and had not previously experienced an osteoporotic fracture (99.3%, n = 111,471).Furthermore, most patients (91.7%, n = 102,919) had not previously received treatment with antiosteoporotic medications.The majority of patients arrived at the hospital for an unplanned emergency visit (45.1%, n = 50,654) or an unplanned nonemergency visit (36.2%, n = 40,677); 18.4% (n = 20,703) of patients had planned their hospital visit.Overall, the most common comorbid health conditions observed at the index hospitalization visit were hypertension (39.0%, n = 43,795), dementia (14.9%, n = 16,691), diabetes without complications (14.0%, n = 15,667), cerebral vascular disease (11.8%, n = 13,290), and congestive heart failure (10.0%, n = 11,278).The proportions of incident diagnoses for common comorbidities at the index hospitalization were 9.3% (n = 10,452) for cerebrovascular disease, 7.5% (n = 8454) for congestive heart failure, 1.4% (n = 1616) for peripheral vascular disease, and 1.3% (n = 1510) for myocardial infraction (Online Resource 1, Table S Patients' ability to engage in ADLs was assessed during their hospitalization (Table 2), with 88.6 to 97.8% of patients having data available across the different ADL categories.For patients with hip fracture, grooming, using the toilet, walking at ground level or on stairs, and bathing were the ADLs that required the most assistance: 60.1% (n = 33,974) needed assistance with grooming, 33.8% (n = 19,094) needed significant help with using the toilet, 39.5% (n = 22,362) needed significant help with ground-level walking, 52.2% (n = 29,555) were unable to walk stairs, and 78.0%(n = 44,130) were unable to bathe independently.Most patients with a fracture to the vertebrae (61.3%, n = 16,343) were also unable to bathe independently, and 31.5% (n = 8414) were unable to walk stairs.Similarly, 42.8% (n = 12,430) of patients with NVNH fractures were unable to bathe independently.

Baseline
Prior to the index hospital admission, most patients (91.7%, n = 102,919) were not receiving any antiosteoporotic treatment.Among the 9356 patients receiving treatment, active vitamin D 3 (51.1%,n = 4778) and bisphosphonates (47.5%, n = 4441) were the most common (Table 4).Among the patients who were taking medication prior to the index hospital visit, the average number of days from the start of treatment to the index hospital visit was approximately 253 days (median, 312 days) across all cohorts.

Follow-up
Upon discharge, 41.5% (n = 46,536) of patients returned to their home, with check-ups scheduled to occur in the same hospital where they were admitted; 34.3% (n = 38,542) of patients were transferred to a different hospital or to a 23 Page 6 of 14 medical care facility (Table 3).After discharge, 21.1% (n = 23,659; 16.6% hip fracture, 34.6% vertebrae fracture, and 17.3% NVNH fractures) of patients received treatment with antiosteoporotic medications (Table 3).Similar to treatment regimens given during the index hospital visit, the three most common medications administered were active vitamin D 3 (56.7%,n = 13,418), bisphosphonates (50.6%, n = 11,978), and teriparatide (16.8%, n = 3963).Treatment initiation for all patients who received treatment after discharge began at a mean of 17.2 days (median, 1 day) after hospital discharge (Table 3).The duration of the first treatment regimen lasted for a mean of 125.7 days (median, 69 days).Patients who discontinued all treatment remained on their initial treatment regimen for a mean 60.3 days (median, 31 days) prior to discontinuing (Table 5).Among the total sample of patients who discontinued (n = 15,109), 43.6% (n = 6593) of patients either discontinued one regimen and then switched to another or discontinued a regimen and then restarted the same regimen at a later date (Table 5).Patients who discontinued treatment without switching discontinued treatment at a mean of 60.3 days (median, 31 days) following treatment initiation.The mean number of days from initiation of the first treatment to the switch or restart of the second treatment was 141.0 days (median, 127 days).There was a mean of 85 days (median, 80 days) between discontinuing one treatment and switching to or restarting another.
Max, maximum; min, minimum; MPR, medication possession ratio; NVNH, non-vertebral non-hip; SD, standard deviation a Includes 11 patients with an MPR of 0.80

Treatment adherence
Among patients who received treatment after the index hospitalization (n = 23,659), 63.9% (n = 15,109) of patients were not adherent to the initial treatment regimen (Table 3).Additionally, 32.4% of patients who received treatment after the index hospitalization had an MPR of ≥ 0.80 (n = 7659).Low adherence rates were also observed when we examined the association of adherence rates with other variables, such as concomitant medication at discharge (i.e., patients with and without concomitant medication) and discharge destination (e.g., home, nursing home, geriatric, or social welfare facility).We found that medication adherence was lower in patients with no concomitant medication at discharge (MPR ≥ 0.8 in 28.9% of patients) compared with patients that had concomitant medication at discharge (MPR ≥ 0.8 in 37.8% of patients).We also observed lower adherence in patients discharged to a nursing home, geriatric facility, or social welfare (MPR ≥ 0.8 in 24.9% of patients) compared with patients discharged to their homes (MPR ≥ 0.8 in 36.1% of patients).Finally, similar MPRs were indicated across the three fracture cohorts, with a mean MPR of 0.4 (SD, 0.4) for patients with hip fracture, 0.4 (SD, 0.4) for patients with vertebrae fracture, and 0.5 (SD, 0.4) for patients with NVNH fractures.

Treatment receipt at follow-up
Additional variables were examined for their association with receipt of treatment after the index hospitalization visit

Discussion
This large retrospective review of patient medical records was conducted to better understand the current treatment patterns of patients hospitalized for osteoporotic fracture in Japan along with the variables associated with receipt of treatment during hospitalization and after discharge.We found that hip fracture was the most common type of hospitalized fracture among both men and women.Most patients who were hospitalized for an osteoporotic fracture initiated their index hospital visit in an unplanned fashion and did not have a previous diagnosis of osteoporosis or a previous fracture.The 0.72% of patients in our study with a previous osteoporotic fracture diagnosis is substantially lower than the 41.5% prevalence rate of multiple fractures in patients with osteoporosis and aged ≥ 50 years that was reported in the 2012-2014 Japan National Health and Wellness Survey [19].
The reason for the discrepancy in rates is unclear but could be partially due to the use of a medical claims database which allowed for patient follow-up only at a single institution (rather than across care settings) versus patient self-report in a survey.Over one-half of the patients in each fracture cohort could perform ADLs, such as eating and transferring, independently or with minor assistance, but patients with hip fracture needed more assistance with grooming, using the toilet, and walking at ground level or on stairs, and patients from all three cohorts needed assistance with bathing.Most patients were not taking antiosteoporotic medications prior to their index hospital visit; and while in the hospital, 25.5% of the total patient sample received antiosteoporotic medications for treatment of their fracture.The most common treatment regimens for fracture were active vitamin D 3 , bisphosphonates, and teriparatide, and these medications were sometimes given in combination.Calcitonin was another medication prescribed for pain relief from fracture during the hospital visit and to a lesser extent upon discharge.The treatment patterns observed in this study are broadly aligned with the results of other studies conducted across different countries (United States [20,21]; China, Hong Kong, Singapore, South Korea, Malaysia, Taiwan, and Thailand [22]; Korea [21]; Spain [21]; and the UK [23]), which found that bisphosphonates were the most commonly used drug category after osteoporotic fracture.Additionally, studies conducted in India [24], Italy [25], and France [26] observed that vitamin D 3 was frequently prescribed to patients after fracture, although the literature regarding vitamin D 3 supplementation patterns post-fracture is limited.Moreover, in agreement with our findings, bisphosphonates and active vitamin D 3 were the most commonly prescribed treatments after hip and vertebral fracture in a recent analysis of health insurance claims in Japan [27].In the current study, approximately 21% of the total patient sample received treatment with antiosteoporotic medications upon discharge, which was similar to the rate of 18.7% observed in a previous chart review study of female patients treated at 25 hospitals in Japan [6].In our analysis, we found that several variables contributed to the likelihood of patients receiving treatment after discharge: patients who were older than 59 years, who were female, who had higher CCI scores, and who received three or more medications upon discharge were more likely to receive treatment after discharge.Although treatment rate was higher in patients who were discharged to their home rather than to a nursing home or geriatric care center, adherence was generally low, regardless of discharge destination.
The literature regarding osteoporosis treatment patterns and adherence outcomes suggests insufficient adherence to antiosteoporotic medications [18,28,29].In the current study, we found that only 32.4% of patients who received treatment after the index hospitalization were adherent to their medication (MPR ≥ 0.80) during the 12-month observation period.In contrast to our findings, Nakatoh et al. [18] observed higher medication adherence over a 2-year follow-up period in Japanese patients with osteoporosis (MPR ≥ 0.80 in 49.6% of patients), although the authors concluded that these adherence rates remained inadequate.Similar to our observation that medication adherence is higher in patients with concomitant medication at discharge, Nakatoh et al. [18] found that medication adherence during the 2-year follow-up period was moderately higher in patients with polypharmacy compared to patients without polypharmacy.A population-based cohort study in Taiwan [29] examined the association of adherence to antiosteoporotic agents with mortality risk in older adults with hip fractures, using data from the Taiwan National Health Insurance Database.Results showed that individuals with higher medication adherence rates had lower mortality rates at 1, 3, and 5 years following the hip fracture than individuals with lower treatment adherence [29].These results underscore the importance of treatment adherence in achieving longer-term benefits.Furthermore, a systematic literature review in 2019 was conducted to understand the condition-related factors for nonadherence with antiosteoporotic medications, including motivation of patients, side effects of medication, perceived benefits, and forgetfulness [28].The authors acknowledged the challenge of low treatment compliance and offered solutions for improving adherence, including patient education, dosing simplification, and flexibility [28].Overall, treatment with antiosteoporotic medications can improve health outcomes, yet adherence is a common challenge [28,29].
One limitation of this study is the possibility that patients had osteoporosis treatment for several years prior to the captured baseline period for the current study.Because we had 1 year of baseline data available on which to base results, the current results may overestimate the rate of patients who did not have prior osteoporosis treatment.Another limitation was that only data within the medical records were available for analysis, as there was no communication with patients or healthcare providers to seek additional information; therefore, results may not be generalizable to all patients who were hospitalized for osteoporotic fracture.For example, although approximately 50% of the patients were discharged to their home, another 34% were discharged to another hospital or medical facility.After discharge to another facility, it may be possible that these patients received additional nonpharmacological care such as physical therapy, and hence the patients did not adhere to the initial regimen.Furthermore, the use of a hospital-based database allowed for patient follow-up only at a single institution, which may have limited the information available for patients that sought care across multiple settings.Use of the MDV database may also limit generalizability to the general Japanese population, as it contains data for hospitalized patients and outpatients in need of healthcare services.However, the database is suitable for our population of interest, as fracture patients in Japan are typically referred to acutecare hospitals [30].A strength was the large sample size of available patient data for abstraction and analysis.The MDV database contained much relevant data for examining treatment patterns from multiple hospitals in Japan and allowed for a longitudinal assessment of variables.Furthermore, the retrospective design of this study allowed for the examination of real-world patient data, which supports generalizability of results.

Conclusion
This study provides an updated treatment landscape for clinical characteristics and treatment patterns of older individuals in Japan who were hospitalized for treatment of osteoporotic fracture.Despite osteoporotic fracture being a major health concern in older Japanese populations, both treatment with antiosteoporotic medications and adherence to treatment regimens remain generally low.

Table 2
Activities of daily living scores during index hospital visit

Table 3
Treatment rates and adherence

Table 4
Treatment regimens from baseline to follow-up after discharge Note: treatment group percentages are calculated out of the total number of treated patients.Subgroup percentages are calculated as a percentage of the associated treatment group