6.2 T-Scores, Z-Scores, and Reference Database Selection

6.2 T-Scores, Z-Scores, and Reference Database Selection

Quick Answer: Both scores express a patient's BMD in standard deviations, and they differ only in what they compare against. The T-score uses a young-adult reference mean and SD; the Z-score uses an age-matched reference. T-scores drive WHO diagnostic classification in postmenopausal women and men 50 and older. Z-scores are the correct score for premenopausal women, men under 50, and children — where a value of −2.0 or lower is reported as "below the expected range for age."

The Two Formulas

$\text{T-score} = \frac{\text{Patient BMD} - \text{Young-Adult Mean BMD}}{\text{Young-Adult SD}}$

$\text{Z-score} = \frac{\text{Patient BMD} - \text{Age-Matched Mean BMD}}{\text{Age-Matched SD}}$

The numerator is the patient's distance from the reference center. The denominator converts that distance into standard-deviation units so that values from different sites and different scanners are expressed on a common scale.

Worked Example

A 62-year-old woman has a femoral neck BMD of 0.652 g/cm². The reference database gives a young-adult mean of 0.858 g/cm² with SD 0.120, and an age-matched mean for 62-year-old women of 0.736 g/cm² with SD 0.118.

$\text{T-score} = \frac{0.652 - 0.858}{0.120} = \frac{-0.206}{0.120} = -1.72$

$\text{Z-score} = \frac{0.652 - 0.736}{0.118} = \frac{-0.084}{0.118} = -0.71$

Both numbers describe the same bone. The T-score of −1.7 places her in the osteopenic range against young adults; the Z-score of −0.7 says she is close to typical for her age. Nothing is contradictory — age-related bone loss is normal, and "normal for 62" is still a level of bone mass associated with fracture risk. That is precisely why the T-score, not the Z-score, drives diagnosis in this population.

Which Score for Which Patient

Reporting Z-Scores

The ISCD terminology is specific and appears on exams verbatim:

• Z-score of −2.0 or lower: report as "below the expected range for age."
• Z-score above −2.0: report as "within the expected range for age."

Notice what is absent: the words osteoporosis and osteopenia. Those terms belong to WHO T-score classification and are not applied on the basis of a Z-score. A premenopausal woman with a Z-score of −2.4 is not "osteoporotic" on densitometry alone; she has bone mass below the expected range for her age, which prompts a search for a secondary cause. Diagnosing osteoporosis in that patient requires clinical criteria such as a fragility fracture or a documented secondary cause.

A low Z-score at any age is a signal to investigate secondary causes — hyperparathyroidism, malabsorption, hypogonadism, glucocorticoid exposure, myeloma — because it means the patient has lost more bone than her age explains.

Reference Database Selection

The reference database is the hidden variable in every score. Change it and the T-score changes while the measured BMD does not.

• Femoral neck and total hip: the NHANES III database is the accepted reference standard in the United States, and ISCD recommends a uniform Caucasian (non-race-adjusted) female reference for computing T-scores at these sites in both women and men.
• Lumbar spine: manufacturers supply their own young-adult reference data, which is one reason spine T-scores are less standardized across systems than hip T-scores.
• Sex-matching for men: ISCD's United States position applies a white female reference for hip T-scores in men. Some manufacturers offer male reference options; consistency within a facility matters more than the choice, and the reference used should be stated in the report.
• Software and database updates change T-scores without any change in the patient. This is a core reason serial comparison uses absolute BMD in g/cm², never T-scores.

The practical technologist rules are short: enter demographics accurately, apply the facility's chosen reference database consistently, and never silently switch reference options between a baseline and a follow-up study.

Which Sites Are Used for Diagnosis

Not every region the software reports is diagnostically valid. ISCD guidance:

The lowest of the valid sites governs classification. If the PA spine gives −1.8, the femoral neck −2.6, and the total hip −2.3, the diagnostic T-score is −2.6 and the classification is osteoporosis.

Ward's area deserves its own note because it is the trap. It is a small, low-density triangular region defined by the intersection of trabecular bundles in the femoral neck. It reliably produces the lowest T-score on the report and it has poor precision, so using it would over-diagnose osteoporosis and would monitor poorly. The software displays it; the report does not classify from it.

Percent of Young Adult and Percent of Age-Matched

Some reports also show percent-of-reference values:

$\text{Percent of young adult} = \frac{\text{Patient BMD}}{\text{Young-Adult Mean}} \times 100$

In the worked example above, $0.652 \div 0.858 \times 100 = 76%$ of the young-adult mean. These percentages are descriptive only. Diagnostic classification uses the T-score, not the percentage, because a given percentage corresponds to different numbers of standard deviations at different sites depending on the reference SD.

Common Errors

• Applying a T-score to a 34-year-old woman and calling her osteopenic.
• Using the Z-score to diagnose osteoporosis in a 70-year-old — her Z-score may be normal precisely because her peers have also lost bone.
• Classifying from Ward's area because it produced the lowest number.
• Comparing T-scores between a baseline and a follow-up after a software or database update.
• Averaging T-scores across sites rather than taking the lowest.