Signal Evaluation: Bisphosphonates and Atypical Femoral Fractures
Introduction
This association is a useful pharmacovigilance case study because the signal developed through several complementary evidence streams rather than one decisive experiment.
Clinicians recognised unusual low-trauma fractures with a characteristic radiographic pattern. Case reports and series suggested a relationship with longer antiresorptive exposure. Epidemiological studies quantified association, while analyses of duration and discontinuation added temporal evidence. Biological reasoning supplied plausibility, and regulators translated the evidence into product-information and risk-management action.
The conclusion was not that every femoral fracture in a person taking a bisphosphonate is drug-related, nor that osteoporosis treatment should be stopped automatically. It is more precise: atypical femoral fractures are rare, clinically distinctive events associated with bisphosphonate exposure, especially longer exposure, and the individual benefit-risk decision must include the competing risk of osteoporotic fracture.
Recognising the signal
Case reports and phenotype
Atypical femoral fractures are generally described as low-trauma fractures in the subtrochanteric or femoral-shaft region with characteristic radiographic features. Thigh or groin pain may precede a complete fracture; incomplete and bilateral lesions can occur.
The phenotype matters. Case review should seek radiographs or specialist confirmation where feasible, fracture location, trauma mechanism, prodromal symptoms, contralateral findings, treatment history, glucocorticoid exposure and relevant comorbidity.
A case report is a signal input, not a denominator. It can show that an event exists and generate a hypothesis, but cannot estimate incidence or establish causality on its own.
Why phenotype specificity mattered
The signal became more credible because the fractures were unusual and clinically coherent. A specific case definition can reduce outcome misclassification and make an association easier to study.
If definitions change between reports and studies, the apparent strength of association can change. A serious assessment should define the outcome, distinguish complete from incomplete lesions where relevant, and explain confirmation.
Epidemiological evidence
Duration and discontinuation
Observational studies reported increased risk associated with bisphosphonate use and a stronger association with longer duration in several analyses. The 2011 Swedish nationwide study reported a substantial adjusted association; later work showed rising risk with longer exposure and declining risk after treatment cessation.
Exact estimates depend on design, population, outcome definition and comparator. The durable lesson is the pattern across studies: duration can modify risk, and a fall after discontinuation supports—but does not prove—causality.
The 2020 Kaiser Permanente analysis by Black and colleagues found increasing hazard with longer exposure and a rapid decrease after discontinuation. It also reported that absolute atypical-fracture risk remained low compared with reductions in hip and other osteoporotic fractures in the studied populations.
Absolute versus relative risk
Relative measures compare groups; absolute risk supports decisions. A high relative risk can correspond to a rare event. Conversely, a rare serious event matters greatly to an individual with prodromal pain or an incomplete fracture.
Interpret the population, comparator, exposure duration, outcome definition, absolute rate where available, confidence interval, residual uncertainty and fractures prevented by treatment. Do not transport one estimate directly to every bisphosphonate, dose, indication or patient.
Biological plausibility and alternative explanations
Remodelling suppression
Bisphosphonates inhibit osteoclast-mediated bone resorption and have durable effects on bone turnover. A plausible mechanism is that prolonged suppression impairs repair of ordinary microdamage in mechanically stressed regions. This helps explain the phenotype and duration pattern, but is not standalone proof.
Bias and confounding
Treated and untreated patients differ in age, fracture risk, comorbidity, glucocorticoid exposure, mobility, body size and healthcare contact. Detection, treatment-selection and outcome-classification bias can affect estimates.
A credible evaluation combines designs and asks whether the association is consistent, phenotype-specific, duration-related, reduced after discontinuation, biologically coherent and robust to alternatives.
Regulatory assessment and risk management
Product information
European regulators reviewed case reports, literature, epidemiology and company data. European product information and safety communications describe atypical femoral fractures as rare, often associated with longer treatment, and advise attention to thigh or groin pain and evaluation of the opposite femur when appropriate.
The wording supports recognition and periodic reassessment; it does not provide a fixed “safe” duration for every person.
Clinical evaluation
For an individual with suspected atypical fracture, treatment decisions require clinical assessment. Discontinuation may be considered while the fracture is evaluated, but the decision must account for osteoporotic-fracture risk and alternatives. This article is not individual medical advice.
At population level, risk minimisation should preserve prevention of common osteoporotic fractures while improving recognition of the rare event. That is benefit-risk reasoning.
What this teaches about signal management
- Detect an unusual and coherent event.
- Validate the phenotype and improve case completeness.
- Search for additional cases and background data.
- Quantify association with appropriate designs.
- Test duration, discontinuation, class and subgroup patterns.
- Examine plausibility and competing explanations.
- Assess absolute risk and treatment benefit together.
- Translate the conclusion into proportionate action.
- Continue monitoring after the action.
The strongest conclusion is not “the drug caused every event”. It is a calibrated statement about association, conditions, uncertainty and action.
References
- Schilcher et al., Risk of atypical femoral fracture during and after bisphosphonate use
- Black et al., Atypical Femur Fracture Risk versus Fragility Fracture Prevention
- Lo et al., Risk of complete atypical femoral fracture with oral bisphosphonate exposure
- EMA, CHMP September 2010 monthly report
- MHRA, Bisphosphonates: atypical femoral fractures
- EMA, GVP Module IX: Signal management
Regulatory Note
This case study does not establish a universal treatment duration, class-wide incidence or individual clinical management. Consult current product information and applicable national or regional guidance for a specific product or patient.