DEXA Bone Density: What This Test Actually Measures

How DEXA Quantifies Bone Density

DEXA works by passing two low-dose X-ray beams at different energy levels through bone. The scanner measures how much radiation each beam loses as it passes through tissue. Bone absorbs more energy than soft tissue, and the difference between the two beams allows the machine to isolate bone mineral content from surrounding fat and muscle. The result is bone mineral density (BMD), expressed in grams per square centimeter (g/cm²).

This dual-energy technique was first validated in the late 1980s and became the WHO reference standard for diagnosing osteoporosis in 1994 [1]. The scanner produces a two-dimensional projection image, not a volumetric measurement. That means DEXA reports areal BMD rather than true volumetric density. Larger bones can appear denser simply because the X-ray path travels through more mineralized tissue. This distinction matters clinically: a tall patient with large vertebrae may register a higher BMD than a shorter patient with identical volumetric density [2].

Radiation exposure is minimal. A standard DEXA scan delivers 1 to 10 microsieverts, roughly one-tenth the dose of a standard chest radiograph [3]. The exam takes 10 to 20 minutes, requires no contrast injection, and needs no fasting. Patients lie flat on an open table while the scanner arm passes over the measurement sites.

The Three Skeletal Sites and Why Each Matters

DEXA measures BMD at three standard locations: the lumbar spine (L1 through L4), the femoral neck, and the total proximal hip. Each site responds differently to aging, hormonal changes, and pharmacotherapy.

The lumbar spine is the most metabolically active site. It contains a high proportion of trabecular (spongy) bone, which turns over faster than cortical bone. That makes the spine the earliest site to show bone loss after menopause and the most responsive site during bisphosphonate therapy. The FIT trial (N=2,027) demonstrated that alendronate increased lumbar spine BMD by 8.8% over 3 years versus placebo [4]. Spine BMD can be falsely elevated by degenerative disc disease, vertebral compression fractures, or aortic calcification, all of which add radiodense material to the scan field [5].

The femoral neck is a small region of the proximal femur where the bone transitions from trabecular to cortical. The WHO recommends using the femoral neck T-score as the reference site for fracture risk classification and FRAX calculations [1]. Hip fracture risk doubles with each standard deviation decrease in femoral neck BMD [6].

The total hip reading averages BMD across a broader region, making it less susceptible to positioning artifacts than the femoral neck alone. The International Society for Clinical Densitometry (ISCD) recommends using the lowest T-score among these three sites for diagnostic classification [7].

T-Scores: The Diagnostic Backbone

A T-score compares a patient's BMD to the mean BMD of a healthy young adult reference population (typically a 30-year-old white female from the NHANES III database). The score is expressed in standard deviations above or below that reference mean.

The WHO diagnostic thresholds, established in 1994 and reaffirmed by the ISCD, are:

• Normal: T-score of −1.0 or higher
• Osteopenia (low bone mass): T-score between −1.0 and −2.5
• Osteoporosis: T-score of −2.5 or lower
• Severe osteoporosis: T-score of −2.5 or lower with one or more fragility fractures [1]

These cutoffs were derived from epidemiological data showing that approximately 30% of postmenopausal white women meet the osteopenia threshold and 16% meet the osteoporosis threshold [8]. The thresholds were calibrated so that the prevalence of "osteoporosis by T-score" would approximate the lifetime fracture risk in this population.

T-scores apply specifically to postmenopausal women and men aged 50 and older. Using T-scores in premenopausal women or younger men can lead to overdiagnosis because peak bone mass varies widely and these populations have different fracture epidemiology [7].

Z-Scores: When Age-Matching Matters

Z-scores compare a patient's BMD to the expected mean for someone of the same age, sex, and ethnicity. A Z-score of −2.0 or lower is classified as "below expected range for age" by the ISCD [7].

Z-scores are the preferred metric for three populations: premenopausal women, men under 50, and children. In these groups, a low Z-score suggests that something beyond normal aging is eroding bone. The ISCD 2019 Official Positions state: "In premenopausal women and men younger than 50, the diagnosis of osteoporosis should not be made on the basis of BMD alone" [7]. A Z-score of −2.0 or below in a 35-year-old woman, for example, should trigger a workup for secondary causes such as hyperparathyroidism, celiac disease, hyperthyroidism, or glucocorticoid excess.

The distinction between T-scores and Z-scores is not cosmetic. A 40-year-old woman with a femoral neck T-score of −2.6 might have a Z-score of only −1.2 if she is compared to age-matched peers, because average BMD at that site has already declined from the peak. Labeling her "osteoporotic" based on the T-score alone could lead to inappropriate bisphosphonate therapy in a patient whose bone density is actually within the expected range for her age [9].

Precision, Least Significant Change, and Serial Monitoring

DEXA is not a perfect ruler. Every scan carries a small measurement error called precision error, typically 1% to 2% coefficient of variation (CV) at the lumbar spine and 1.5% to 2.5% at the femoral neck [10]. This precision error defines the least significant change (LSC): the smallest BMD change that exceeds measurement noise.

For most clinical DEXA systems, the LSC at the spine is approximately 0.033 g/cm² (roughly a 3% change). A BMD shift smaller than the LSC cannot be confidently attributed to true biological change versus scanner variability.

The Endocrine Society's 2019 Clinical Practice Guideline on postmenopausal osteoporosis recommends serial DEXA monitoring every 1 to 2 years for patients on pharmacotherapy, using the same scanner and the same technologist when possible to minimize repositioning error [11]. Dr. Clifford Rosen, writing in the New England Journal of Medicine, noted: "Changes in bone density of 3 to 5 percent at the spine over 2 years generally reflect a true treatment effect, while smaller shifts often fall within the noise of the measurement" [12].

Switching between GE Lunar and Hologic scanners introduces systematic BMD differences of 5% to 10% that cannot be cross-calibrated reliably at the individual patient level [10]. Patients should return to the same facility for follow-up scans.

How DEXA Results Guide Clinical Decisions

A single DEXA scan generates a BMD number and a T-score. Those values do not exist in isolation. Clinicians integrate DEXA results with the FRAX algorithm (Fracture Risk Assessment Tool), clinical risk factors, and sometimes vertebral fracture assessment (VFA) to make treatment decisions.

The USPSTF recommends osteoporosis screening with DEXA for all women aged 65 and older and for younger postmenopausal women whose 10-year fracture risk (calculated by FRAX using clinical factors alone, without BMD) equals or exceeds that of a 65-year-old white woman with no additional risk factors [13]. This threshold corresponds to a 10-year major osteoporotic fracture risk of approximately 9.3%. For men, the USPSTF concluded in 2018 that evidence is insufficient to recommend routine screening, though the Endocrine Society recommends screening men aged 70 and older [11].

Treatment thresholds vary by guideline. The National Osteoporosis Foundation (now the Bone Health and Osteoporosis Foundation) recommends pharmacotherapy when the hip T-score is −2.5 or below, or when the T-score falls between −1.0 and −2.5 with a FRAX 10-year probability of 3% or more for hip fracture or 20% or more for major osteoporotic fracture [14]. The AACE 2020 guidelines use similar thresholds but add that patients with T-scores of −2.5 or lower at any measured site, or those with fragility fractures regardless of T-score, should be considered for therapy [15].

Dr. E. Michael Lewiecki, director of the New Mexico Clinical Research and Osteoporosis Center, has stated: "DEXA is the gatekeeper for osteoporosis treatment, but the T-score alone is an incomplete measure of fracture risk. Integrating BMD with FRAX and clinical context is essential for appropriate treatment decisions" [16].

What DEXA Does Not Measure

DEXA has blind spots. It does not assess bone quality, which encompasses microarchitecture, collagen cross-linking, mineralization homogeneity, and microdamage accumulation. Two patients with identical T-scores can have markedly different fracture risks because their trabecular microstructure differs [17].

Trabecular bone score (TBS) is a texture analysis applied to the lumbar spine DEXA image that partially addresses this gap. TBS correlates with trabecular connectivity and has been integrated into FRAX as an adjustment factor since 2015 [18]. A low TBS (below 1.230) modestly increases the FRAX-calculated fracture probability independent of BMD.

DEXA also cannot detect cortical porosity at the femoral neck, a feature that quantitative CT (QCT) and high-resolution peripheral QCT (HR-pQCT) can evaluate. These volumetric techniques measure true three-dimensional density in mg/cm³ and distinguish cortical from trabecular compartments, but they deliver higher radiation doses and are not included in WHO diagnostic criteria [2].

Vertebral fracture assessment (VFA), a low-dose lateral spine image acquired on the DEXA scanner, can identify prevalent vertebral compression fractures that would otherwise go undetected. The ISCD recommends VFA for postmenopausal women with T-scores below −1.0, adults with height loss of 4 cm or more, or patients with suspected but undocumented vertebral fractures [7]. Approximately two-thirds of vertebral fractures are clinically silent, and their presence independently increases future fracture risk regardless of BMD [19].

Factors That Falsely Alter DEXA Results

Several conditions produce artifactually high or low BMD readings that can mislead clinicians if not recognized.

Falsely elevated BMD: Degenerative disc disease and osteophytes at the lumbar spine are the most common cause of spuriously high readings in older adults. Aortic calcification, vertebral compression fractures, and surgical hardware also add radiodense material to the scan field [5]. When spine BMD is unreliable due to artifacts, clinicians should rely on the hip sites for diagnosis.

Falsely low BMD: Significant weight loss (more than 10% of body weight) can reduce the soft-tissue baseline that DEXA uses for calibration, potentially underestimating true BMD. Laminectomy at the lumbar spine removes posterior elements and can decrease the projected BMD of affected vertebrae [20].

Body composition also affects accuracy. In patients with BMI above 35, DEXA precision degrades because the X-ray beams attenuate more through thick soft tissue, and some patients exceed the scanner's weight limit (typically 130 to 160 kg depending on the model) [10].

The ISCD recommends that densitometrists review each scan image for artifacts and exclude affected vertebrae from the analysis. At minimum, two evaluable lumbar vertebrae must remain for a valid spine BMD result [7].

Raising and Lowering Bone Density: What DEXA Tracks Over Time

DEXA is the standard tool for monitoring treatment response. The anabolic agent teriparatide (Forteo) increases lumbar spine BMD by approximately 9.7% and femoral neck BMD by 2.8% over 18 months, as demonstrated in the key Neer trial (N=1,637) [21]. Romosozumab (Evenity) produced a 13.3% increase in lumbar spine BMD at 12 months in the FRAME trial (N=7,180), the largest gain observed with any single osteoporosis agent [22].

Antiresorptive agents produce more modest but sustained gains. Denosumab (Prolia) increased lumbar spine BMD continuously over 10 years in the FREEDOM Extension trial, reaching a cumulative gain of 21.7% from baseline [23]. Bisphosphonates typically produce 5% to 8% gains at the spine over 3 years, with smaller increments at the hip [4].

Lifestyle interventions produce measurable but smaller effects. A meta-analysis of 43 randomized trials (N=4,320) found that resistance exercise increased lumbar spine BMD by 1.0% to 1.5% compared to controls, while calcium plus vitamin D supplementation reduced bone loss without producing net gains [24].

Declining BMD on serial DEXA while on therapy raises the question of treatment failure or poor adherence. The ISCD defines inadequate treatment response as a BMD decline exceeding the LSC or occurrence of a new fracture after 1 to 2 years of appropriate pharmacotherapy [7]. This threshold applies only when the patient has been adherent. Oral bisphosphonate adherence drops below 50% by 12 months in real-world studies [25].