A DXA body-fat Z-score flagged twice as many athletes with absent periods as BMI did

Body mass index was never designed for athletes. It cannot distinguish muscle from fat, and female athletes typically carry greater lean mass than the general population from which BMI thresholds were derived [s1]. That is a known problem in principle. A study in the British Journal of Sports Medicine measures how much it costs in practice, using menstrual status as the outcome [s1].

What was compared

Researchers analysed 388 female athletes aged 15–30 competing at Tier 2 or above who had dual-energy X-ray absorptiometry scans, classifying them from clinical records as amenorrhoeic (n=159), oligomenorrhoeic (n=84) or naturally menstruating (n=145) [s1].

They then asked whether a DXA-derived per cent fat Z-score could detect menstrual status, and whether a Z-score cut-off below −1.0 performed better than the standard risk thresholds in use: BMI for athletes over 20, and per cent expected body weight for those aged 20 and under [s1].

Per cent fat Z-score was superior to the traditional BMI or per cent expected body weight thresholds in discriminating between menstrual status groups [s1].

The trade the cut-off makes

The sensitivity gain is large. Using a per cent fat Z-score cut-off below −1.0 improved sensitivity (p<0.0001) for predicting amenorrhoea to 68.9% and for the combined oligomenorrhoea/amenorrhoea group to 57.7%, against 29.3% and 25.9% respectively for the traditional BMI or per cent expected body weight cut-offs [s1].

Those traditional numbers are the finding that should give clinicians pause. A threshold that identifies fewer than three in ten athletes with amenorrhoea [s1] is not a screening tool that happens to underperform; it is one that misses most of what it is looking for in this population.

The cost is specificity, and the paper reports it plainly. The Z-score cut-off did not improve specificity for predicting amenorrhoea (75.4%) or the combined group (79.9%) compared with the traditional cut-offs (83.9%, p=0.0078; and 85.2%, p=0.14) [s1]. The amenorrhoea comparison is statistically significant; the combined-group comparison is not.

So the new cut-off finds roughly twice as many affected athletes and produces more false positives while doing it. Whether that trade is worth making depends on what follows a positive result — an argument the study frames but does not settle.

The authors' conclusion is about treatment targets rather than screening: when attempting to resume normal menstruation, achieving a per cent fat Z-score of −1.0 or above, rather than using BMI or per cent expected body weight targets, may be a better goal for athletes with REDs-related amenorrhoea or oligomenorrhoea [s1].

What the design cannot establish

This is a cross-sectional comparison of diagnostic thresholds, not a trial. It shows that a measure discriminates between groups defined by clinical records; it does not show that moving an athlete's Z-score above −1.0 restores menses. The authors phrase the recommendation as "may be a better goal" [s1], and the difference between a discriminating marker and a validated treatment target is exactly the gap that phrasing acknowledges.

Menstrual status came from clinical records [s1], which means classification depends on what was documented rather than on prospective cycle tracking. Hormonal contraception, which alters bleeding patterns without indicating energy status, is a standing complication in this literature and is not addressed in the study summary.

The sample is also selected: 388 athletes at Tier 2 or above who had a DXA scan [s1]. Athletes referred for DXA are not a random sample of athletes, and the prevalence of amenorrhoea in this group — 159 of 388 — is far higher than in the athletic population at large, which inflates positive predictive value relative to a general screening setting.

Why the threshold question matters clinically

The reason to care about detecting low energy availability early is what it does to bone, and a retrospective analysis published in October gives that consequence a measured shape.

Researchers analysed 82 elite athletes, 30.5% female, mean age 23.4 ± 7.6 years, presenting to an outpatient clinic, diagnosing relative energy deficiency in sport using the IOC REDs Clinical Assessment Tool Version 2 [s2]. REDs was diagnosed in 24% of the athletes, and stress fractures were observed far more frequently in those with REDs than without: 70% versus 25% (p<0.001) [s2].

The bone biochemistry pointed the same way. Osteocalcin and procollagen type 1 N-terminal propeptide were reduced in athletes with REDs compared with strength-based athletes (p<0.01), while urinary deoxypyridinoline/creatinine and calcium excretion were elevated (p<0.05), indicating suppressed bone formation alongside increased bone resorption [s2]. Athletes with REDs showed significantly reduced Z-scores at the lumbar spine and hip compared with strength and endurance athletes without REDs (p<0.05) [s2], and high-resolution peripheral quantitative computed tomography revealed lower bone volume to tissue volume and trabecular bone mineral density at the distal radius and tibia, with more pronounced effects at the load-bearing tibia (p<0.01) [s2].

That study is retrospective, single-clinic, and drawn from athletes who presented for assessment [s2] — the same selection issue, in a smaller sample. It does not establish that the bone changes are caused by energy deficiency rather than accompanying it.

The through-line

Two papers with similar limitations point at the same operational problem. Athletes with the condition have measurable skeletal deterioration and far higher stress fracture rates [s2], and the thresholds most commonly used to flag them identify under 30% of the amenorrhoeic cases in a DXA-scanned athlete cohort [s1]. A better-calibrated marker does not fix that on its own, but it makes the miss rate visible, which is the first step.

Sources

• Dual-energy X-ray absorptiometry per cent fat Z-score as a predictor of menstrual status in adolescent and young adult female athletes, British Journal of Sports Medicine, published online 10 March 2026
• Impact of Relative Energy Deficiency in Sport (REDs) on Bone Health in Elite Athletes: A Retrospective Analysis, Journal of Cachexia, Sarcopenia and Muscle, October 2025