A T-Score Is Not a Fracture Forecast
The subject of this series is how osteoporosis and fracture risk are assessed in real clinical decisions. Its first question is simple but easy to miss: why can a bone-density result matter without being enough to define a person’s chance of breaking a bone? Understanding that distinction gives readers a more useful way to interpret a T-score, distinguish screening from diagnosis and treatment, and recognize when a number needs clinical context rather than automatic action.
A T-score is memorable because it appears to turn a complicated biological problem into one figure. A result of −2.5 or lower is widely recognized as the densitometric definition of osteoporosis. But the outcome that matters to a person is not the label on the scan. It is whether a fragile bone fractures, what happens afterward, and whether an intervention can reduce that harm. The number is evidence about one part of that problem. It is not the problem in its entirety.
The result people remember is not the outcome they fear
A fragility fracture is a fracture caused by a fall from standing height or less—an event that would not fracture most healthy people. Major osteoporotic fractures include fractures of the hip, spine, wrist and shoulder. These are not minor changes in a report. Osteoporotic fractures can bring psychological distress, another fracture, loss of independence, and difficulty performing ordinary activities. The consequences are especially serious after a hip fracture: the US Preventive Services Task Force reports that only 40% to 60% of people who experience one recover their prefracture mobility and ability to perform daily activities.
That is why fracture prevention is the practical endpoint of osteoporosis assessment. A diagnostic category can be useful because it organizes evidence and helps clinicians decide what to do. Yet a category is not the same as an outcome. If assessment focuses only on whether a person crosses a density threshold, it can lose sight of people who fracture without meeting that threshold and of the other factors that make a fracture more or less likely.
The T-score is a comparison. It describes how a person’s bone mineral density, or BMD, at a measured site compares with the average BMD of a reference population of young, healthy White women. The World Health Organization definition uses a value of −2.5 standard deviations or lower at the hip or lumbar spine. That definition became a reference standard for males and females and for people of all racial and ethnic groups. It gives clinicians a common language for describing low bone density.
A common language is valuable. It is not a crystal ball. The same recommendation that describes the T-score also says that advancing age is a stronger determinant of fracture than bone density alone. Older adults can have much higher fracture rates than younger adults with the same BMD because declining bone quality and a rising chance of falling add to the danger. The threshold classifies density; it does not absorb every reason a bone may fail or every circumstance in which a person may fall.
The distinction becomes even more important when we look at where fractures occur in a population. The USPSTF notes that most fragility fractures occur in people without densitometric osteoporosis—that is, people whose DXA T-scores are above −2.5. This does not mean that a low T-score is harmless or that the threshold should be discarded. It means that a threshold can identify an important group without containing the whole fracture burden.
A density number is a signal, not the whole risk picture
The scan behind the T-score is central DXA, or dual-energy X-ray absorptiometry. It measures BMD at central skeletal sites such as the total hip, femoral neck or lumbar spine. The USPSTF describes central DXA as the most commonly used bone measurement test for screening. It correlates with bone strength and clinical fracture outcomes, uses a low dose of radiation, and predicts risk at a particular site best when density is measured at that site. It is also the established standard used to determine T-scores and was used to determine eligibility in nearly all trials of bone-conserving medicines.
Those facts establish why DXA matters. They do not turn BMD into a complete forecast. A measurement can correlate with an outcome while leaving substantial variation between individuals. In the evidence reviewed by the USPSTF, cohorts using BMD alone to predict major osteoporotic fractures reported areas under the receiver operating characteristic curve, or AUCs, from 0.60 to 0.80. For hip fractures, the reported range was 0.64 to 0.86. In one study of women aged 45 to 54 years, femoral-neck BMD had an AUC of 0.64.
An AUC is a way to describe discrimination across a group: how often a measure tends to rank people who experience an outcome above those who do not. It is not an individual’s probability, and it does not say that a particular T-score guarantees or rules out a fracture. The ranges also vary by population, outcome, follow-up and study design. A useful predictor can therefore leave two people with similar results on different sides of the eventual outcome.
The diagram’s direction matters. The T-score is not shown as the final box because it is not the final question. It enters a chain that also requires information about the person and the decision being made. The chain is not a claim that every factor can be reduced to one perfect composite number. It is a reminder that the purpose of measurement is to improve a prevention decision.
This is also why the same T-score can carry different practical meaning. Consider two hypothetical people with the same femoral-neck result. If one is younger and has no history of falling, while the other is older and has recurrent falls, the density comparison has not changed—but the surrounding fracture risk has. The example does not allow us to assign either person a personal percentage from the available evidence. It does show why the result cannot be interpreted as if age and falling were irrelevant background details.
The timing of follow-up reveals the same principle. Cohort studies summarized by the USPSTF found that repeating BMD at four- to eight-year intervals did not add accuracy to fracture prediction. Transition to densitometric osteoporosis also varied with baseline T-score and age: one analysis estimated almost 17 years for 10% of women with normal BMD to develop osteoporosis, compared with about five years for women whose baseline T-score was between −1.50 and −1.99. A baseline density result can inform how quickly a change might deserve attention, but it does not create a universal retesting schedule for every person.
Three questions hide inside a single report
A bone-density report can appear to answer one question while clinical care is actually asking three. Screening asks whom to test among people who do not already have known osteoporosis or a history of fragility fracture. Diagnosis asks how to classify the measured density. Treatment asks whether a person’s expected benefit from an intervention justifies its burdens and possible harms. These questions are connected, but evidence for one does not automatically settle the others.
The 2025 USPSTF recommendation makes that separation visible. It recommends screening women aged 65 years or older and postmenopausal women younger than 65 who are at increased risk after clinical assessment. For men, it gives an I statement: current evidence is insufficient to determine the balance of benefits and harms. The recommendation applies to adults aged 40 years or older without known osteoporosis or a fragility-fracture history, and it excludes people whose bone loss is secondary to specified medical conditions or chronic use of medications such as glucocorticoids.
For a younger postmenopausal woman, the suggested route is not “find a T-score first and decide everything from it.” It begins by identifying risk factors, then uses a clinical risk assessment tool to help decide whether DXA is warranted. Screening can consist of central DXA with or without fracture-risk assessment. If osteoporosis is found, the pathway still has work to do: the person should be further evaluated, counseled and, when appropriate, offered evidence-based management. Detection without a reasonable downstream response cannot deliver the full benefit that screening is meant to provide.
The Canadian quick-reference guide offers another concrete view of the same logic. It asks clinicians to assess risks for low BMD, fractures and falls; it varies BMD-testing indications by age and risk factors; and it directs pharmacotherapy decisions toward a 10-year fracture-risk calculation using BMD with FRAX preferred or CAROC. The details and scope differ from the USPSTF approach, but the underlying lesson is similar: BMD is placed inside a sequence of clinical judgments rather than treated as a self-interpreting verdict.
The first conclusion of this series is therefore not that T-scores are unreliable. It is more precise: a T-score answers a density question, while fracture prevention asks an outcome question. The number can be central to diagnosis and to the evidence behind treatment, yet it cannot by itself account for age, bone quality, falling, prior fracture, clinical conditions, medication exposure, model assumptions or the evidence available for a particular population.
Before deciding what to do with a T-score, we need to understand exactly how the measurement is made and what its statistical performance can—and cannot—establish. That is the foundation for using the number carefully rather than asking it to carry a decision it was never designed to carry.