Bone mineral density scanning + 10-year fracture probability modeling for osteoporosis screening
DEXA has been the gold-standard technique for measuring bone mineral density (BMD) since the early 1990s, replacing single-photon and quantitative CT methods with a faster, lower-dose, more reproducible scan. It underpins nearly every major osteoporosis guideline worldwide, including the WHO diagnostic criteria and the FRAX fracture prediction tool.
DEXA scanners generate an X-ray beam alternating (or filtered) between two distinct photon energy levels, typically ~40 keV and ~70 keV. Bone and soft tissue attenuate these two energies differently — bone (high calcium content) strongly attenuates low-energy photons relative to high-energy ones, while soft tissue attenuates both more uniformly.
By measuring the differential attenuation ratio pixel-by-pixel as the scanner arm sweeps across the hip and lumbar spine, software algorithms subtract out the soft-tissue signal and isolate bone mineral content (BMC, grams of hydroxyapatite) within each pixel. Dividing BMC by the projected bone area yields areal bone mineral density (BMD, g/cm²) — the value reported clinically.
Standard scan sites are the femoral neck and total hip (predicts hip fracture best) and the lumbar spine L1–L4 (predicts vertebral fracture and is more sensitive to early change, but confounded by osteophytes and aortic calcification in older adults).
DEXA measures areal density (2D projection), not true volumetric density — a systematic limitation that under-estimates BMD in small-boned individuals and over-estimates it in large-boned individuals. Volumetric QCT avoids this but at far higher radiation dose and cost.
Major guideline bodies (USPSTF, NOF, ISCD, WHO) recommend routine BMD screening for:
• All women aged 65 and older, regardless of risk factors • All men aged 70 and older • Postmenopausal women under 65 and men 50–69 with clinical risk factors (low body weight, prior fracture, glucocorticoid use, family history, smoking) • Anyone with a fragility fracture after age 50 • Patients on long-term glucocorticoid therapy (≥3 months at ≥5 mg/day prednisone-equivalent), regardless of age
Repeat scanning intervals are typically 2 years for patients near a treatment threshold, and up to 10–15 years for those with normal baseline BMD and few risk factors, per the extended screening-interval studies (Gourlay et al., NEJM 2012).
Modern DEXA software renders the raw attenuation data as a color- or grayscale-coded density map overlaid on the skeletal image, letting technologists and radiologists visually identify regions of low density (dark/red) versus preserved density (bright/blue) before quantitative region-of-interest analysis is applied.
The scanner output feeds two numbers per site: the T-score (comparison to young-adult reference) and the Z-score (comparison to age-matched peers) — the T-score drives diagnosis and treatment thresholds, while the Z-score helps flag secondary causes of bone loss when unexpectedly low for age.
The T-score converts a patient's raw BMD measurement into a standardized position on the population bell curve of peak bone mass, established from a young, healthy reference cohort. The World Health Organization's 1994 diagnostic thresholds — built directly from T-score cutoffs — remain the global standard for defining osteoporosis today.
In 1994, a WHO Study Group defined four diagnostic categories based on T-score, measured at the femoral neck using DEXA:
• Normal: T-score ≥ -1.0 • Osteopenia (low bone mass): T-score between -1.0 and -2.5 • Osteoporosis: T-score ≤ -2.5 • Severe (established) osteoporosis: T-score ≤ -2.5 with one or more fragility fractures already present
These cutoffs were chosen so that roughly 30% of postmenopausal white women meet the osteoporosis definition — the group in which the original fracture-epidemiology data were gathered — and the framework has since been extended, imperfectly, to men and other populations.
A T-score is not the same as a Z-score. T-score compares to peak young-adult bone mass (diagnostic use); Z-score compares to age-, sex-, and ethnicity-matched peers (used to flag atypical bone loss, e.g. Z ≤ -2.0 suggests investigating secondary osteoporosis causes).
Raw BMD in g/cm² varies enormously by scanner manufacturer, calibration, and skeletal site, making cross-study and cross-clinic comparison unreliable. Expressing density as a standard-deviation distance from a defined reference population normalizes for this variability and creates a portable, interpretable clinical statistic.
Epidemiological cohort studies (e.g. the Rotterdam Study, Study of Osteoporotic Fractures) established that fracture risk rises roughly exponentially, not linearly, as T-score falls — each one-SD reduction in femoral neck BMD is associated with an approximate doubling of hip fracture risk, which is why FRAX uses BMD as a continuous multiplicative input rather than a simple threshold flag.
T-score is a strong predictor but an incomplete one: a substantial fraction of fragility fractures occur in patients with osteopenia rather than osteoporosis, simply because far more people fall into the larger osteopenic population. This observation — formalized by Kanis and colleagues at the University of Sheffield — was the direct motivation for building FRAX: a tool that combines BMD with independent clinical risk factors to estimate absolute fracture probability, rather than relying on a density threshold alone.
Developed by the WHO Collaborating Centre at the University of Sheffield and launched in 2008, FRAX integrates femoral neck BMD with independent, partially BMD-independent clinical risk factors to compute an individualized 10-year probability of fracture — moving osteoporosis management from a density threshold to an absolute-risk framework.
FRAX combines femoral neck BMD (or T-score, optional) with clinical risk factors that each carry independent predictive weight beyond bone density:
• Age (40–90 years) — risk rises steeply with age independent of BMD • Sex — separate risk curves for men and women • Body mass index (BMI) — low BMI independently raises risk • Previous fragility fracture — prior fracture roughly doubles future fracture risk • Parental history of hip fracture — heritable component of bone fragility • Current smoking — impairs osteoblast function and estrogen metabolism • Long-term oral glucocorticoid use — suppresses osteoblasts, increases osteoclast activity • Rheumatoid arthritis — independent inflammatory risk factor • Alcohol intake ≥3 units/day — dose-dependent skeletal toxicity • Secondary osteoporosis causes (when BMD unavailable) — e.g. type 1 diabetes, hyperthyroidism, hypogonadism, chronic liver disease
Roughly half of FRAX's risk factors act largely independently of BMD — a patient with normal bone density but a prior fragility fracture and glucocorticoid use can still cross the pharmacologic treatment threshold, which is why FRAX outperforms BMD-only screening at identifying who truly benefits from therapy.
FRAX was derived from meta-analysis of prospective, population-based cohorts spanning Europe, North America, Asia, and Australia — totaling roughly one million patient-years of observation. Poisson regression models related each risk factor, and their interactions, to observed fracture and mortality hazard, then integrated over the remaining lifetime to yield a 10-year absolute probability (competing mortality risk is explicitly modeled, since a patient must survive to fracture).
Because background fracture incidence and mortality vary substantially by country, FRAX is calibrated separately for each nation/region using local epidemiologic data — an identical set of risk factors can output meaningfully different percentages in different national models.
Femoral neck BMD is the preferred skeletal input to FRAX because it has the largest and most consistent evidence base linking it to hip fracture across diverse populations — lumbar spine BMD is not used by the algorithm because vertebral degenerative change makes it a noisier predictor in older adults. FRAX can still be calculated without any BMD value (using clinical risk factors alone), but adding femoral neck T-score materially improves discrimination, especially in patients near the treatment threshold.
FRAX outputs two headline percentages: the 10-year probability of a major osteoporotic fracture (clinical spine, forearm, hip, or proximal humerus) and the 10-year probability of hip fracture specifically. These absolute-risk numbers, not the T-score alone, drive treatment decisions under NOF and USPSTF guidance.
Major osteoporotic fracture risk aggregates four fracture sites with heterogeneous consequences — vertebral and hip fractures carry far higher morbidity and mortality than forearm or humeral fractures, but combining them yields a more statistically stable, more common outcome to predict. Hip fracture risk is reported separately because it is the single most clinically consequential fracture type — associated with the highest mortality, disability, and cost — and some populations (e.g. very elderly patients) can have major-fracture risk below threshold while hip risk alone justifies treatment.
The U.S. National Osteoporosis Foundation (NOF, now the Bone Health & Osteoporosis Foundation) issued cost-effectiveness-derived treatment thresholds in 2008, subsequently adopted by USPSTF guidance:
• Treat if 10-year major osteoporotic fracture risk ≥ 20% • Treat if 10-year hip fracture risk ≥ 3% • Treat regardless of FRAX score if T-score ≤ -2.5 (osteoporosis by densitometric definition alone) • Treat regardless of FRAX score after a hip or vertebral fragility fracture (very high risk, treatment nearly always indicated)
These thresholds were calibrated so that the cost per quality-adjusted life-year (QALY) gained from generic bisphosphonate therapy remains below commonly accepted willingness-to-pay benchmarks — they are explicitly a health-economic, not purely biological, cutoff.
A patient with osteopenia (T-score -1.8) but a prior wrist fracture, a parent with hip fracture, and current smoking can easily cross the 20%/3% thresholds despite never meeting the densitometric definition of osteoporosis — this is precisely the population FRAX was designed to capture.
In clinical FRAX reports, both percentages are typically displayed as simple numeric outputs alongside a graphical scale; some implementations add color-coded risk bands (green/amber/red) to aid rapid interpretation by non-specialist providers. The gauge visualization in this stage mirrors that convention — filling toward red as risk crosses the NOF/USPSTF thresholds — while the underlying computation always remains a continuous probability, not a discrete category.
Once absolute fracture risk is established, management branches into two pathways: universal lifestyle and fall-prevention measures for everyone, and pharmacologic therapy layered on top once risk exceeds the intervention threshold. The choice among pharmacologic agents depends on fracture site, severity, renal function, and treatment duration limits.
Regardless of computed fracture risk, foundational measures are recommended for essentially all patients being evaluated for bone health:
• Adequate calcium intake (~1200 mg/day for postmenopausal women and men over 70), preferentially from diet • Vitamin D sufficiency (800–1000 IU/day, titrated to serum 25-hydroxyvitamin D ≥30 ng/mL) • Weight-bearing and resistance exercise, shown to modestly increase or preserve BMD and — more importantly — improve muscle strength and balance • Smoking cessation and alcohol moderation • Home fall-hazard assessment, vision correction, and gait/balance training, particularly in elderly patients where falls (not bone fragility alone) precipitate most hip fractures
When 10-year major fracture risk ≥20% or hip risk ≥3% (or T-score ≤-2.5, or a prior fragility fracture), pharmacologic therapy is added:
• Bisphosphonates (alendronate, risedronate, zoledronic acid, ibandronate) — first-line for most patients; inhibit osteoclast-mediated resorption; reduce vertebral fracture risk ~40–50% and hip fracture risk ~40–50% in high-risk populations; oral weekly/monthly or annual IV zoledronic acid • Denosumab — subcutaneous RANKL inhibitor dosed every 6 months; potent antiresorptive, useful in renal impairment where bisphosphonates are contraindicated; rebound bone loss and vertebral fracture risk if abruptly discontinued without bisphosphonate transition • Anabolic agents (teriparatide, abaloparatide, romosozumab) — reserved for very high-risk patients (T-score ≤-3.0, multiple fractures, or failure of antiresorptive therapy); stimulate new bone formation rather than only slowing resorption; teriparatide/abaloparatide limited to 2 years lifetime use; romosozumab combines anabolic and antiresorptive action but carries a cardiovascular safety signal
Anabolic-first strategies (teriparatide or romosozumab followed by an antiresorptive) produce larger BMD gains and greater fracture-risk reduction than antiresorptive-first sequencing in very high-risk patients — a sequencing principle established by trials such as VERO and ARCH and now embedded in Endocrine Society guidance.
Bisphosphonate therapy is typically reassessed after 3–5 years ("drug holiday" consideration) given evidence that antifracture benefit persists after discontinuation for a period, weighed against rare but serious long-duration risks (atypical femoral fracture, osteonecrosis of the jaw) that rise with cumulative exposure. Denosumab, in contrast, has no drug-holiday option — resorption rebounds rapidly upon discontinuation and must be followed by a bisphosphonate to prevent rebound vertebral fractures. Repeat DEXA scanning at 1–2 year intervals monitors treatment response and guides these transitions.