A T-score isn’t a measurement. It’s a comparison — and the thing you’re being compared to is a specific group of people, chosen decades ago, who may not resemble you at all.
That sounds like a technicality. It isn’t. It changes what the number on your report means, and it’s the part almost no one explains at the appointment.
What a T-score compares
Start with the basic definition, then we’ll get to the complication.
Your T-score expresses how far your bone density sits from the average of young adults at peak bone mass — not from people your own age. The unit is standard deviations. A T-score of −2.0 means you’re two standard deviations below that young-adult average.
The WHO classification most reports use:
- −1.0 or above — normal
- Between −1.0 and −2.5 — low bone mass
- −2.5 or below — osteoporosis
The ISCD notes that while “osteopenia” is still used for that middle band, “low bone mass” or “low bone density” is preferred[2]. Their reasoning is worth quoting in spirit: people with low bone mass are not necessarily at high fracture risk[2].
The reference population
Now the part that surprised me.
The WHO international reference standard for diagnosing osteoporosis is a T-score of −2.5 or less at the femoral neck. And the database that score is calculated from is specified precisely: female, white, aged 20–29, from NHANES III[1].
NHANES III was a US health survey conducted between 1988 and 1994.
So a T-score calculated in Seoul, Singapore or São Paulo is, at the hip, referenced against young white American women measured over thirty years ago. That specification has been carried forward across successive editions of the ISCD positions[4].
This is deliberate rather than accidental. The ISCD position is explicit: use a uniform white, non-race-adjusted female reference for women of all ethnic groups, and the same reference for men of all ethnic groups[1]. Where local reference data exist, they should be used to calculate Z-scores but not T-scores[1].
Why this matters if you’re Asian
Peak bone density differs between populations, and the difference isn’t trivial.
Research has increasingly reported that peak bone mineral density differs between Asian and Caucasian populations, with some of the gap attenuating after adjusting for body size or weight[3]. A 2002 study found Singaporean men had 10% lower peak bone density at the lumbar spine and 5% lower at the femoral neck compared with a Caucasian reference population[3].
Set that beside the diagnostic rule. If your peak was lower to begin with, and you’re measured against a reference built from a population with a higher peak, your T-score starts from a disadvantage that has nothing to do with your current bone health trajectory.
There’s a counterpoint that keeps the picture honest. The patterns of bone loss between races appear to be the same[3] — what differs is the starting point, not the rate of decline.
This is why the same source notes that recommendations may vary according to local requirements[1]. The uniform reference is a standardisation choice, not a claim that populations are identical.
Hip and spine use different reference databases
Here’s a second wrinkle that affects how you read your own report.
The ISCD position states that manufacturers should continue using NHANES III data as the reference standard for femoral neck and total hip T-scores — but should continue using their own databases for the lumbar spine[1].
| Site | T-score reference | Implication |
| Femoral neck | NHANES III (standardised) | Comparable across machines |
| Total hip | NHANES III (standardised) | Comparable across machines |
| Lumbar spine | Manufacturer’s own database | May vary by machine |
[Table 1] Reference databases by measurement site · Sources: ISCD Official Positions[1]
That means a spine T-score from one scanner and a spine T-score from another are not guaranteed to be on the same scale. Hip values are standardised; spine values are not.
The practical consequence: if you’re tracking change over time, being scanned on the same machine matters more than it might appear — and it matters most for the spine.
What T-scores can’t be applied to
One more limit worth knowing, because it comes up with screening devices.
T-scores from measurements other than DXA at the femoral neck, total femur, lumbar spine or one-third radius cannot be used under the WHO diagnostic classification[2]. Those include heel ultrasound and peripheral devices sometimes offered at health fairs or pharmacies.
Such devices can be useful for identifying who might benefit from a proper scan. They don’t produce a number that the diagnostic thresholds apply to.
Where Z-scores come in
If T-scores compare you to young adults, Z-scores compare you to people your own age and sex.
The ISCD position is that where local reference data are available, they should be used for Z-scores[1]. So a Z-score can be population-appropriate in a way a T-score deliberately isn’t.
Z-scores are the primary measure in premenopausal women, men under 50, and children and adolescents — groups for whom “compared to peak young-adult density” isn’t a meaningful frame.
If your report shows both, it’s worth asking which one your clinician used to reach their assessment.
Notes for readers residing in Korea
Bone density screening is included in Korea’s national health screening programme at certain ages for women, and the eligible ages have been expanded in recent years — worth checking the current criteria rather than an older guide.
Two things follow from this post specifically.
First, if your report was generated in Korea, the hip T-score is still referenced against the NHANES III database. That’s the international standard, not a local quirk.
Second, if you’re comparing a Korean scan to one done in another country, ask whether both were on the same manufacturer’s equipment — particularly for the spine.
Closing
The T-score is a useful number, and none of this suggests otherwise. Standardising the reference database is what makes diagnosis consistent across clinics and countries — that consistency has real value.
What I’d assumed was that the comparison group was somehow neutral. It isn’t neutral; it’s specified. Young white American women, surveyed between 1988 and 1994, at the hip. For the spine, whatever database your scanner’s manufacturer supplies.
Knowing that doesn’t change your bone density. It changes how much weight to put on a single number crossing a single threshold — which, as the ISCD itself notes, doesn’t map neatly onto fracture risk anyway.
Next in this series: calcium absorption by salt form, and why the timing advice differs between them.
At a Glance
- A T-score compares you to young adults at peak bone mass, not to people your age
- WHO bands: −1.0 or above normal, −1.0 to −2.5 low bone mass, −2.5 or below osteoporosis
- The reference database is specified as female, white, aged 20–29, NHANES III — a US survey from 1988–1994[1]
- ISCD recommends this uniform reference for all ethnic groups, with local data used only for Z-scores[1]
- Asian populations have been reported with lower peak bone density — around 10% lower at the spine, 5% at the femoral neck in one Singaporean study[3] — though patterns of loss appear similar
- Hip T-scores use a standardised database; spine T-scores use the manufacturer’s own — so spine values may not be comparable across machines[1]
- WHO thresholds do not apply to heel ultrasound or other peripheral devices[2]
- ISCD prefers “low bone mass” to “osteopenia,” noting such people are not necessarily at high fracture risk[2]
※ This article explains how bone density reports are constructed and is for general information only. It does not replace medical diagnosis, interpretation or treatment. Reference standards and screening criteria are revised periodically. Please have your results interpreted by a clinician, and do not start, stop or adjust medication or supplements based on a number you have read yourself.
References
- International Society for Clinical Densitometry, “Official Positions — Adult” (2023), https://iscd.org/official-positions-2023/
- ISCD Official Positions — Adult (2007 edition, terminology and device limitations), https://www.iscd.org/official-positions/2007-iscd-official-positions-adult/
- “Osteoporosis in men—East and West: Can the twain meet? A perspective from Asia”, ScienceDirect, https://www.sciencedirect.com/science/article/pii/S2405525524001201
- 2015 Adult Official Positions of the ISCD (reference standard specification), https://www.aub.edu.lb/fm/CaMOP/Documents/iscd-adult-official-positions.pdf