Understanding your DEXA results: T-scores and Z-scores explained

A DEXA report arrives with several numbers on it, and the two that matter most are the T-score and the Z-score. They are both comparisons, and they compare you with different people.

The T-score

The T-score compares your bone density with the average peak bone density of a healthy young adult of the same sex. It is expressed in standard deviations, so a T-score of −1.0 means your bone density sits one standard deviation below that young-adult average.

The World Health Organisation categories, as applied in current UK guidance, are ¹:

Two points are worth understanding about this scale.

First, it is a continuum that has had lines drawn on it. A T-score of −2.4 and a T-score of −2.6 are almost identical measurements that fall into two differently named categories.

Second, the osteopenia band is very wide. Someone at −1.1 and someone at −2.4 are both described as having osteopenia, and their situations are not comparable.

The Z-score

The Z-score compares your bone density with what would be expected for someone of your own age, sex and body size. Because bone density declines with age, an older person can have a low T-score and a perfectly ordinary Z-score. That combination usually reflects normal age-related loss.

A low Z-score — generally taken as below −2.0 — is different ¹. It means your bone density is lower than expected for your age group, which raises the possibility of a specific underlying cause rather than the gradual decline that comes with time. That typically prompts investigation for causes such as thyroid overactivity, hyperparathyroidism, coeliac disease, low testosterone, or the effects of medication.

In children, adolescents, premenopausal women and men under 50, the Z-score is used rather than the T-score, and the T-score categories above do not apply.

Why sites can disagree

Your report will give scores for the lumbar spine and for the hip, and sometimes for the forearm. They frequently differ. The spine is more responsive to change and often shows a lower score early on. It is also more prone to artificially high readings in older people, because arthritic change, calcification in the aorta, and previous vertebral fractures all add mineral to the image without adding strength.

The hip is less affected by these artefacts and is the more reliable site for predicting hip fracture specifically. Where the two disagree substantially, the lower of the two is generally used for classification, but the interpretation belongs with a clinician who can see the images.

Why the score is not the same as your risk

Bone density is one input into fracture risk, not the whole of it. Fracture happens when a load exceeds what a bone can withstand, which depends on the strength of the bone and on how it is loaded. Risk calculators such as FRAX combine bone density with age, previous fracture, parental hip fracture, smoking, alcohol, corticosteroid use, rheumatoid arthritis and body mass index. These tools drive clinical decisions more directly than the T-score does ^{1, 2}.

Trabecular bone score adds a further dimension by assessing the internal architecture of the bone rather than its density, and identifies some people whose risk is higher than their T-score implies ³. Falls risk matters too, and is often the more modifiable half of the equation.

Comparing scans over time

A single scan is a position. Two scans are a direction, and the direction is usually more informative. For that comparison to mean anything, the scans need to be genuinely comparable. Differences between scanner models and between machines of the same model are large enough to obscure real change, so repeat scanning on the same machine is strongly preferable. Where that is not possible, the reporting radiologist needs to know.

Change also has to exceed the measurement precision before it counts as change. Small movements between scans are frequently noise rather than signal, which is one reason scans are not usually repeated more often than every two years for bone density purposes.