2026 年 72 巻 3 号 p. 233-243
Objective : To investigate the relationship between physical activity levels and atypical femoral fractures (AFF) in bisphosphonate (BP) users.
Methods : In the Assessing Daily-Activity-Level for Decisions on Bisphosphonate Prescription (ADD-BP) study, we retrospectively analyzed 27 women with AFF from five hospitals (2019-2023), categorized into BP (n = 18) and no-BP (n = 9) groups. Pre-injury physical activity was assessed via mobility aid usage, Long-Term Care Insurance System status, and independence in daily living.
Results : Median ages were 77 years (BP) and 86 years (no-BP). In the BP vs. no-BP groups, 83.3% vs. 33.3% used no mobility aids, 11.1% vs. 22.2% used canes, and 5.6% vs. 33.3% used walkers. No BP users (0%) vs. 33.0% of no-BP users were housebound ; others were independent.
Conclusion : BP users with AFF exhibited higher physical activity levels than non-users, suggesting low activity may not elevate AFF risk in long-term BP use. These findings support the safety of extended BP therapy for fragility fracture prevention in less active patients. Future case-control studies are needed to confirm these results. This study was approved by the Fukushima Medical University Ethics Committee (REC2023-176).
Bisphosphonates (BPs) are widely prescribed for the treatment of osteoporosis, effectively preventing bone density loss and fractures) 1) . However, long-term use of BPs has been associated with the side effect of atypical femoral fractures (AFF) 2 , 3) . Consequently, concerns about the risk of AFF have led many physicians to hesitate in prescribing BPs, resulting in patients not receiving adequate treatment for sufficient durations 4) . Particularly in patients with low physical activity levels, fractures can cause a further decline in physical function, potentially necessitating intensive therapeutic interventions5). Given their cost-effectiveness, safety, and efficacy, BPs are essential in developing treatment programs for these low-activity patients.
The mechanism by which BPs induce AFF involves a severe suppression of bone turnover, which accelerates the accumulation of microdamage and increases bone fragility 6) . Additionally, physical activity is associated with mechanical stress on bones and plays a role in bone metabolism 7) . We hypothesized that, in BP users, physical activity levels are closely related to mechanical loading on the femur and the accumulation of microdamage, potentially influencing AFF risk. If this hypothesis holds, lower physical activity levels may correlate with a reduced risk of AFF ( Fig. 1), providing a rationale for recommending long-term BP use in patients with low physical activity levels.
Therefore, we designed the Assessing Daily-Activity-Level for Decisions on Bisphosphonate Prescription (ADD-BP) study. The ADD-BP study comprises two phases : a preliminary study and the main case-control study. This paper reports the results of the preliminary study, which aimed to investigate pre-injury levels of activities of daily living (ADL) among patients with AFF. The preliminary study was designed to assess the feasibility of the main study and to explore background information regarding the history of BP use and pre-injury physical activity levels in AFF patients.
The Assessing Daily-Activity-Level for Decisions on Bisphosphonate Prescription study examines the influence of physical activity levels on the incidence of atypical femoral fractures (AFF) in addition to the use of bisphosphonates *→ : indicates acceleration. **⊣ : indicates inhibition. ⁺Mobility aids : a cane, walker, and wheelchair. ☨SBBT : severely suppressed bone turnover.
Data were accessed for research purposes on March 8, 2024. Authors had access to information that could identify individual participants during data collection ; however, all data were anonymized prior to analysis to maintain participant confidentiality.
This study was conducted by the principles of the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement 8) and was approved by the Ethics Committee of our institution (REC2023- 176). The study was registered and initiated in the Japan Registry of Clinical Trials ( https://jrct.niph.go.jp/re) (jRCT1020230053). The requirement for patient consent was waived owing to the retrospective nature of the study.
We included patients who were admitted to five major hospitals in the northern region of Fukushima Prefecture from 2019 to 2023 and met the four major diagnostic criteria for AFF by the American Society for Bone and Mineral Research9) based on plain radiographs on admission. The exclusion criteria included 1) patients with a history of denosumab use, 2) patients with a history of contralateral AFF, 3) patients with osteogenesis imperfecta, and 4) patients using BP for metastatic bone tumors.
The included patients were classified into two groups : the no-BP group, patients with no history of bone resorption inhibitor use, and the BP group, which included those with a history of BP use. In the BP group, the duration of BP use prior to injury was recorded.
OutcomesPatient Information at Admission
Patient information collected at admission was age, sex, height, weight, body mass index (BMI), kidney function (estimated glomerular filtration rate [eGFR] from blood tests), and pre-injury medication history, which encompasses blood pressure medications, proton pump inhibitors (PPI), and steroids.
Radiographic Image
Radiographic images were used to evaluate the fracture site, categorized as either subtrochanteric or shaft. Subtrochanteric fractures were defined as those with a fracture line within 5 cm distal to the lesser trochanter, whereas fractures in the diaphyseal region beyond this point were defined as shaft fractures 10) . In addition, the coronal femoral curvature (CFC), sagittal femoral curvature (SFC), and lateral cortical thickness index (LCTI) at the lesser trochanter (LT) and shaft levels were evaluated on the contralateral healthy side 11) .
Pre-injury Physical Activity Level
A history of mobility aid use, the status of the Long-Term Care Insurance System, the independence degree of daily living for older individuals with dementia, and the independence degree of daily living for seniors with disabilities were assessed. Details are provided in “Data Resources and Handling.”
Sample SizeThis present study was a preliminary investigation, and we collected data from patients diagnosed with AFF at participating facilities over the 5 years preceding 2023. Therefore, no sample size calculation was performed in the study.
Data Resources and HandlingAge, sex, height, weight, BMI, and eGFR were collected from medical records at the time of hospitalization. Sex was treated as a categorical variable, whereas the others were considered continuous variables.
Data on patients’ medication history before injury was collected by physicians, nurses, and pharmacists and recorded in the medical records. A history of minodronate, ibandronate, risedronate, alendronate, or zoledronic acid use was considered a history of BP use. PPI use was identified if the patient received daily doses of PPIs. Steroid use was identified if the patient received ≥5 mg of prednisolone daily. All of these were treated as categorical variables. In addition, the duration of BP use was recorded and treated as a continuous variable.
In the radiological evaluation, plain radiographs of the contralateral healthy side were used to determine whether the fracture site was subtrochanteric or shaft. In addition, CFC, SFC, and LCTI (LT level and shaft level) were measured on the contralateral healthy side, as is standard in the literature. All of these measurements were treated as continuous variables.
Data on pre-injury physical activity levels were collected based on medical records. The history of mobility aid usage was investigated ; independent walking was classified as “No,” walking with a cane as “Cane,” walking with a walker as “Walker,” and moving with a wheelchair as “Wheelchair.”
According to the Long-Term Care Insurance System, which is assessed by the Certification Committee for Long-term Care Need, composed of health, medicine, and welfare experts in Japan 12) , patients are classified into “not certified,” “requiring support,” and “requiring long-term care” owing to the need for assistance. Independent patients are classified as “not certified.” In the “requiring support” state, patients need assistance with daily living activities but do not require admission to care facilities. In the “requiring long-term care” state, patients need more intensive care, including admission to care facilities. The “requiring support” state is further categorized into two levels (Levels 1-2), and the “requiring long-term care” state is divided into five levels (Levels 1-5). As the levels increase, the required assistance increases ( Table 1) 13) .
Data on the independence degree of daily living for older individuals with dementia ( Table 2) and the independence degree of daily living for seniors with disabilities ( Table 3) were collected as important information for determining the status of the longterm care insurance system.
All outcomes regarding pre-injury physical activity levels were recorded as categorical variables.
For continuous variables, the results are expressed as the number and percentage of missing data for each item. For categorical variables, the number of missing data is not presented.
* As the levels increase, the required care time also increases.
Continuous variables were presented as medians and interquartile ranges (IQR), and categorical variables were presented as counts and percentages for each category.
For between-group comparisons, the Mann- Whitney U test was performed for continuous variables, and the chi-square test was performed for categorical variables. The statistical software used in this study was IBM SPSS Statistics for Windows Version 29.0. A statistical significance level of < 5% was considered significant.
Using the eligibility criteria, 33 patients with AFF were selected. Six patients were excluded according to the exclusion criteria ; therefore, 27 patients with AFF were included in the study. The flowchart of the participants is shown in Fig. 2.
Patient CharacteristicsThe characteristics of the included patients are shown in Table 4. The median age was 78 years, and all were female. Nine patients had no history of BP use (no-BP group), and 18 patients had a history of BP use (BP group). In the BP group, information on the duration of BP use was available for nine patients, with a median of 7.25 (IQR 5.25-10.5) years.
Inter-group ComparisonSignificant differences were observed in age, CFC, history of mobility aid use, and independence degree of daily living for the seniors with disabilities ( Tables 5a, b).
In the no-BP and BP groups, the median ages were 86 (IQR 79.0-88.0) and 77 years (IQR 73.0- 81.5), respectively ( p=0.006). The median CFC was 6.20° (IQR 5.45-10.55) and 2.83° (IQR 0.93- 4.49), respectively ( p=0.004).
Regarding the history of mobility aid use, the proportion of patients with no mobility aid use was 33.3% vs. 83.3%, using a cane was 22.2% vs. 11.1%, and using a walker was 33.3% vs. 5.6%, respectively, in the no-BP vs. BP groups ( p=0.049).
Regarding the independence degree of daily living for the seniors with disabilities, patients who were judged as A2 in the no-BP and BP groups were 33.0% vs. 0%, respectively, and the rest were independent ( p=0.03).
Of 33 patients with atypical femoral fractures (AFF), 6 were excluded. The remaining 27 patients were included for analysis. The five hospitals marked with an asterisk ( *) are Fukushima Medical University Hospital, Minami Tohoku Fukushima Hospital, Ohara General Hospital, Fujita Public General Hospital, and Fukushima Red Cross Hospital.
IQR ; interquartile range, BMI ; body mass index
*Mann-Whitney U test, p-value < 0.05.
AFF, atypical femoral fracture ; BP, bisphosphonate ; IQR, interquartile range ; BMI, body mass index ; eGFR, estimated glomerular filtration rate ; CFC, coronal femoral curvature ; SFC, sagittal femoral curvature ; LCTI, lateral cortical thickness index
*Chi-square test or Fisher’s exact probability test, p-value < 0.05.
AFF, atypical femoral fracture ; BP, bisphosphonate ; N.A., not available ; PPI, proton pump inhibitor
In the present study, a few patients with AFF and low physical activity levels were receiving BPs prior to their injuries. In contrast, among AFF patients who were not receiving BP, there were relatively more older patients with low physical activity levels. This observation suggests a potential relationship between physical activity levels and the occurrence of AFF in patients receiving BPs.
Our results indicate that among AFF patients, those receiving BP were younger and had higher physical activity levels, whereas those not receiving BP were older and had lower physical activity levels. This suggests that the characteristics of patients with AFF differ based on their history of BP use. Notably, there were few patients with low physical activity levels among patients with AFF receiving BP. These findings suggest that mechanical stimuli, such as physical activity, play a more significant role in the development of AFF in patients receiving BP compared to patients not receiving BP, which aligns with our hypothesis.
Conventionally, AFF is believed to be primarily caused by severe suppression of bone turnover (SSBT) 14) . However, recent studies suggest that AFF may result from bone fragility, particularly in older adults 15) . Patients who develop AFF as a result of bone fragility are known to exhibit pronounced femoral bending, similar to that observed in patients not using BP in this study.
In this study, the CFC in the BP group was significantly smaller than that in the non-BP group. This finding suggests that in older non-BP users, pronounced femoral bowing increases mechanical stress, while bone fragility reduces resistance to this stress, contributing to AFF. In contrast, in BP users, typically younger, smaller CFCs reduce anatomical mechanical stress ; however, high physical activity levels likely increase mechanical stress on the femur, contributing to AFF occurrence. This is consistent with our hypothesis that higher physical activity levels are associated with an increased AFF risk in BP users.
In summary, in patients with AFF with a history of BP use, both SSBT and physical activity levels affected AFF occurrence, whereas in those without a history of BP use, bone fragility was involved in the development of AFF
BPs and the Prevention of Fragility FracturesBP is an effective drug for the prevention of fragility fractures caused by osteoporosis 1) . However, the long-term use of BP increases the risk of AFF, leading to hesitation among some clinicians to prescribe BP 16 , 17) . However, the incidence of AFF is very low compared to femoral fragility fractures ; therefore, various guidelines recommend the use of BP for a sufficient period in patients with osteoporosis 18 , 19) . Specifically, even after 5 years of oral BP use or 3 years of injectable BP use, long-term use of up to 10 years is permissible to prevent fracture if the fracture risk is high 19 , 20) .
Among patients with AFF receiving BP in our study, the number of AFF cases was small among patients with low physical activity levels. This result suggests potential evidence for long-term BP use in patients with low physical activity levels ; however, confirmation in future studies is warranted.
Challenges of Physicians in Prescribing BPsIn clinical practice, BP may be discontinued before the completion of a sufficient treatment period, based on the judgment of individual physicians 16 , 17) . It is a concern that, in addition to the discontinuation of BP, subsequent interventions for osteoporosis may never be resumed. One reason for the lack of long-term BP prescriptions is that physicians misunderstand or overestimate the risk of the side effects of BP, including AFF. Khosla et al. reported that many physicians strongly believe that the effectiveness of osteoporosis treatment is lower than the risk of side effects ; the study reported that the physicians’ belief had a significant impact on patient adherence 16) . Donato et al. reported that more than half of the patients discontin- ued BP before positive clinical effects were observed. Notably, the most common reason for discontinuation was a recommendation by another doctor to discontinue BP 17) . It is important not only to improve patient compliance with medication but also to provide clinicians with adequate information and guidance to ensure they prescribe BP for an appropriate duration.
To ensure long-term continuation of BP and improve patient adherence, it is essential to address the negative perceptions many physicians have regarding the side effects of BP, particularly AFF. The ADD-BP study could play a crucial role in dispelling these concerns and potentially eliminating the negative perception.
Finally, BP users with higher physical activity levels, likely relatively younger individuals, may have an increased risk of AFF compared to older patients with lower activity levels. However, at the population level, AFF incidence remains low. In this preliminary study conducted in a medical region of approximately 400,000 people, only a few dozen AFF cases were identified over 5 years. The incidence of fragility fractures, such as hip fractures (58.90 per 10,000 person-years), significantly exceeds that of AFF (1.74 per 10,000 personyears) 2) ; therefore, BP use should remain a treatment option for relatively younger and highly active individuals to prevent fragility fractures. In such cases, strict adherence to defined BP treatment durations (5 years for oral, 3 years for intravenous) and drug holidays (typically 1.5 to 3 years) is recommended to mitigate risks of fragility fractures and AFF ; restarting bisphosphonate therapy is considered based on clinical indicators of increased fracture risk, such as declining bone mineral density, elevated bone turnover markers, or new fragility fractures, though criteria are individualized and not universally standardized 18 , 19) .
AFF and Physical Activity LevelsThe population of the northern region of Fukushima Prefecture makes up a quarter of the total population of Fukushima Prefecture. This study was conducted in five major hospitals in the northern region of Fukushima Prefecture. From the characteristics of the case accumulation facilities, it is believed that the patients of this study guarantee the population representativeness of patients with AFF in this district. Among patients with AFF in the northern region of Fukushima Prefecture over the past 5 years, 6 of 9 nonBP users (66.7%) required mobility aids before injury, indicating lower physical activity levels, whereas in the BP group, only 3 of 18 patients (16.7%) used walking aids, and 15 (83.3%) did not use mobility aids. This contrast suggests that AFF incidence is lower among BP users with low physical activity levels requiring mobility aids, while higher physical activity levels, indicated by the absence of mobility aids, likely contribute to AFF risk in BP users, consistent with our hypothesis
A previous study agrees that AFF occurs less frequently in patients receiving BP with low levels of physical activity. According to Mizutani et al., preoperative physical activity was high among AFF surgery patients, with 12 out of 13 walking without a cane and one using a cane 21) . In BP users, high levels of physical activity may potentially increase the incidence of AFF.
Future Prospects of the ADD-BP StudyThe ADD-BP study is a retrospective study designed to include a case-control study. Patients with a history of BP use who have developed AFF will be considered the case group, and patients with a history of BP use will be the control group.
To investigate the relationship between AFF and physical activity levels, it is necessary to consider various confounding factors, including age. In case-control studies, setting up the control group is the most critical and challenging task 22) . Age is a confounding factor between physical activity levels and the occurrence of AFF 23) . The case group in this study, the hospital-based population, is relatively easy to obtain. However, owing to the biases, there is a high possibility that it may be inappropriate as a control group 24) .
Additionally, while medications such as benzodiazepines are known risk factors for falls and fragility fractures 25) , their role in AFF remains unclear, as they are not consistently identified as risk factors in the literature. In this preliminary study, we did not collect data on these medications, as our focus was on the relationship between BP use and AFF. However, in future studies, collecting data on these medications as background variables could provide valuable insights, particularly when comparing AFF with fragility fractures or exploring potential confounding effects.
The results of this study did not contradict our hypothesis ; however, a well-designed case-control study is necessary to confirm the hypothesis. Future research should focus on optimizing control group selection to minimize biases and include additional variables, such as medication history, to en hance the robustness of the findings.
This study has several limitations. First, there may be selection bias. Patients with AFF who participated in this study represent the population of patients with AFF in the northern region of Fukushima Prefecture. However, we may not have collected data on some patients with AFF. These patients may include those who received conservative treatment on an outpatient basis or did not visit a hospital because they healed naturally. This fact is considered a limitation of hospital-based retrospective studies.
Second, we used existing medical records and samples to allow data collection. When initiating the following case-control study, it is necessary to consider collecting additional information besides the existing data. For example, data on BP use duration, a critical confounding factor, was only available for nine of 18 patients in the BP group. Similarly, data on bone density, an important indicator of bone strength, could not be obtained owing to the retrospective observational design. As a multicenter study, variations in bone densitometry equipment across facilities would likely introduce measurement inconsistencies, complicating data interpretation. In future studies, we plan to use computed tomography-derived Hounsfield units at the time of injury 26) to enhance data availability and accuracy. In addition, steroid use is a significant AFF risk factor, and further investigation of steroid duration is warranted.
Third, all the study participants were Japanese. Therefore, in populations comprising different races, the results must be cautiously interpreted 2 , 27) . This is particularly relevant for populations with significantly different body sizes, where our results may not directly apply. The reproducibility of the results needs to be confirmed in countries outside of Japan.
The results of this preliminary study showed that BP users with lower physical activity levels had a lower incidence of AFF in some regional cities in Japan. If it can be demonstrated that low physical activity levels in BP users lead to a reduced risk of AFF, it can serve as evidence for the long-term prescription of BP to decrease the risk of fragility fractures. Further research will be conducted to explore the relationship between physical activity levels and the occurrence of AFF.
None.
The authors declare no conflicts of interest. A separate Conflict of Interest form will be submitted to f-igaku@fmu.ac.jp.
Conceptualization : T.Y. ; Data curation : T.Y., K.S., T.T., H.A., K.S., Y.S., N.S., K.M. ; Formal analysis : T.Y. ; Investigation : T.Y., Y.K., H. Y. ; Methodology : T.Y. ; Project administration : T. Y. ; Supervision : M.S., Y.M. ; Writing - original draft : T.Y. ; Writing - review & editing : T.K., S.S., M.S., T.N., Y.M. All authors approved the final manuscript.
No funding or research grants were received.
