Fracture Risk Is Not Just About Bone Density: Understanding FRAX

The Limitation of Bone Density Alone

For decades, bone mineral density (BMD) measured by DXA has been the cornerstone of osteoporosis diagnosis. The WHO classification — normal, osteopenia, osteoporosis — is based entirely on T-scores. While BMD is unquestionably an important predictor of fracture, it is far from the only one. In fact, the majority of fragility fractures occur in individuals whose bone density falls in the osteopenic range, not the osteoporotic range.

This paradox is straightforward when you consider the numbers: many more people have osteopenia than osteoporosis. Even though the per-person fracture rate is higher in the osteoporotic group, the larger number of people with osteopenia means they contribute more total fractures to the population. This is why a clinical approach that relies solely on bone density to make treatment decisions will inevitably miss many individuals who are at meaningful fracture risk.

What Is FRAX?

FRAX (Fracture Risk Assessment Tool) was developed by the Centre for Metabolic Bone Diseases at the University of Sheffield, UK, under the direction of Professor John Kanis, and was first released in 2008. It is a computer-based algorithm that integrates clinical risk factors with or without femoral neck BMD to calculate the 10-year probability of a major osteoporotic fracture (hip, spine, forearm, or proximal humerus) and the 10-year probability of a hip fracture specifically.

FRAX models have been calibrated for over 70 countries and territories using country-specific fracture and mortality data. This geographic calibration is important because fracture rates vary substantially by region — hip fracture incidence in Scandinavia, for example, is several times higher than in sub-Saharan Africa, even after adjusting for age and bone density.

The Clinical Risk Factors in FRAX

FRAX incorporates the following risk factors, each of which has been shown in large epidemiological studies to predict fractures independently of bone density:

Age

Fracture risk increases exponentially with age, even at the same T-score. A 70-year-old woman with a T-score of -2.0 has a substantially higher fracture risk than a 50-year-old woman with the same T-score. This reflects age-related declines in bone quality, muscle strength, balance, and the increased frequency and severity of falls with aging.

Sex

Women have a higher lifetime fracture risk than men, primarily due to lower peak bone mass, menopause-related bone loss, and longer life expectancy. However, men are not immune — approximately 20-25% of hip fractures occur in men, and the mortality rate after hip fracture is actually higher in men than in women.

Body Mass Index (BMI)

Low BMI is an independent risk factor for fracture, even after accounting for bone density. This may reflect lower soft-tissue padding over bony prominences, reduced mechanical loading on bone, and nutritional factors. FRAX uses BMI as a continuous variable rather than a simple threshold.

Prior Fragility Fracture

A prior fracture after age 50 from a standing height or less is one of the strongest predictors of future fracture. The risk is particularly elevated in the first two years following a fracture. Vertebral fractures are especially predictive — a woman with an existing vertebral fracture has approximately a 4-5 fold increased risk of a subsequent vertebral fracture and a 2-3 fold increased risk of hip fracture.

Parental History of Hip Fracture

A parental hip fracture (specifically hip, not other fracture types) increases fracture risk independently of bone density. This likely reflects inherited aspects of bone geometry, microarchitecture, and possibly fall propensity that are not captured by DXA.

Current Smoking

Smoking is associated with lower bone density, but it also increases fracture risk beyond what the BMD reduction alone would predict. The mechanisms may include direct toxic effects on osteoblasts, reduced calcium absorption, altered estrogen metabolism, and increased cortisol levels.

Glucocorticoid Use

Oral glucocorticoid use (equivalent to 5 mg or more of prednisolone daily for 3 months or more) is a well-established cause of secondary osteoporosis. Glucocorticoids suppress osteoblast function, increase osteoclast activity, and reduce calcium absorption. The fracture risk from glucocorticoids is partially independent of bone density — fractures can occur at higher T-scores than would be expected in non-glucocorticoid-treated individuals.

Rheumatoid Arthritis

Rheumatoid arthritis (RA) is included as an independent risk factor in FRAX, separate from glucocorticoid use. The chronic inflammatory state of RA directly promotes bone resorption through cytokine pathways, and reduced mobility further contributes to bone loss.

Secondary Osteoporosis

This category encompasses conditions known to cause bone loss, including type 1 diabetes, osteogenesis imperfecta, untreated long-standing hyperthyroidism, hypogonadism, premature menopause (before age 45), chronic malnutrition or malabsorption, and chronic liver disease.

Alcohol Intake

Consumption of 3 or more units of alcohol per day is associated with a dose-dependent increase in fracture risk. Alcohol has direct effects on osteoblast function and also increases fall risk.

How FRAX Is Used in Clinical Practice

FRAX outputs two numbers: the 10-year probability of a major osteoporotic fracture and the 10-year probability of a hip fracture. These probabilities are used to guide treatment decisions, though treatment thresholds vary by country and guideline.

In the United States, the National Osteoporosis Foundation (now the Bone Health and Osteoporosis Foundation) recommends considering pharmacological treatment when:

• The 10-year probability of a major osteoporotic fracture is 20% or greater, or
• The 10-year probability of a hip fracture is 3% or greater

These thresholds were derived from cost-effectiveness analyses specific to the US healthcare system. Other countries use different thresholds — the UK, for example, uses age-dependent intervention thresholds that compare an individual's FRAX probability to the average fracture probability of a woman of the same age who has already had a fracture.

Strengths of FRAX

• Validated in large populations: FRAX was developed using data from prospective cohorts encompassing hundreds of thousands of patient-years across multiple continents.
• Can be used without BMD: In settings where DXA is not available, FRAX can be calculated using clinical risk factors alone, providing a reasonable estimate of fracture probability.
• Country-specific calibration: Unlike a one-size-fits-all approach, FRAX accounts for geographic differences in fracture epidemiology.
• Identifies high-risk patients who would be missed by T-score alone: A 75-year-old woman with a T-score of -1.8, a prior wrist fracture, and a parental hip fracture may have a FRAX score that exceeds the treatment threshold, even though her T-score is in the osteopenic range.

Limitations of FRAX

FRAX is a useful tool, but it is not without important limitations that clinicians and patients should understand:

• Risk factors are entered as yes/no: FRAX does not account for dose-response relationships. For example, a patient on 30 mg of prednisone daily has a much higher fracture risk than someone on 5 mg, but FRAX treats both as "yes" for glucocorticoid use. Similarly, the number and severity of prior fractures are not differentiated — a single distal radius fracture is treated the same as multiple vertebral fractures.
• Fall risk is not included: Falls are the proximate cause of most non-vertebral fractures in older adults, yet FRAX does not incorporate fall history or fall risk assessment. This was a deliberate choice by the developers, who argued that fall-related variables did not consistently improve fracture prediction in the development cohorts. However, in clinical practice, fall risk is an important component of fracture prevention.
• It uses femoral neck BMD only: Even when BMD is included, FRAX uses only the femoral neck T-score, not the spine. In patients where the spine shows more severe bone loss than the hip, FRAX may underestimate risk.
• It may underestimate risk in certain populations: Patients with type 2 diabetes, for instance, often have normal or even elevated BMD but still have increased fracture risk, likely due to changes in bone quality. FRAX does not adequately capture this discordance.
• It does not account for recent trends: FRAX provides a snapshot based on current risk factors. It does not incorporate the rate of bone loss, recent changes in health status, or new medications that might alter fracture trajectory.

FRAX Adjustments and Emerging Refinements

Recognizing some of these limitations, several adjustments to FRAX have been proposed and validated:

• Trabecular bone score (TBS) adjustment: TBS is a textural analysis of the DXA spine image that provides indirect information about trabecular microarchitecture. A TBS adjustment has been incorporated into the FRAX online tool, which can reclassify patients across treatment thresholds.
• Glucocorticoid dose adjustment: Tables are available to adjust FRAX probabilities upward for higher glucocorticoid doses and downward for lower doses.
• Recency of fracture: While not yet incorporated into the standard FRAX calculator, research has shown that fracture risk is highest in the 1-2 years immediately following a fracture, a concept termed "imminent fracture risk." This has clinical implications for the urgency of treatment initiation.

Putting It Together: A Clinical Scenario

Consider two 65-year-old women, both with a femoral neck T-score of -2.0 (osteopenia):

• Patient A: No prior fractures, no family history, non-smoker, no glucocorticoid use. Her FRAX 10-year hip fracture probability might be approximately 2.5%.
• Patient B: Prior vertebral fracture, mother had a hip fracture, current smoker. Her FRAX 10-year hip fracture probability might be approximately 7%.

Both women have the same bone density. But Patient B has nearly three times the fracture risk and would meet treatment thresholds under US guidelines, while Patient A might be appropriately managed with monitoring and lifestyle measures. This is the clinical value of FRAX: it moves the conversation beyond a single number (the T-score) to a more nuanced assessment of individual fracture probability.

What You Should Discuss With Your Clinician

If you have had a DXA scan, particularly one showing osteopenia, ask your healthcare provider whether a FRAX assessment has been performed. Understanding your 10-year fracture probability can help inform decisions about whether monitoring alone is appropriate, or whether pharmacological treatment should be considered. The FRAX tool is freely available online at frax.shef.ac.uk, though interpretation of the results should be done in consultation with a qualified clinician who can account for factors that FRAX does not capture.

Fracture risk assessment is not a one-time event. As your clinical risk factors change — with aging, new fractures, changes in medication, or shifts in health status — your fracture probability changes as well. Periodic reassessment is an important component of long-term bone health management.