Diagnostic performance of lumbar computed tomography Hounsfield unit thresholds for osteoporosis and osteopenia: a systematic review and meta-analysis
Authors' conclusion
Vertebral HU strongly discriminate normal from low BMD, providing clinically applicable pooled thresholds for opportunistic CT-based screening and perioperative risk assessment.
Study context
- Population
- Studies reporting vertebral Hounsfield units (HU) referenced to DXA or quantitative computed tomography (CT); databases searched to September 2025
- Participants
- 5,016 participants20 studies
- Comparison
- Vertebral HU on routine CT vs DXA or quantitative CT as reference
- Outcome
- Unified HU cut-offs for identifying osteopenia and osteoporosis
Methodological detailsbivariate random-effects models, reconstructed 2 × 2 analyses
- Bivariate random-effects models
- Reconstructed 2 × 2 analyses
Pooled result
As reported in the abstract
Overall diagnostic performance of vertebral HU
- AUC
- 0.865
- Sensitivity
- 0.80
- Specificity
- 0.83
- Pooled osteopenia threshold, balanced
- 138.9 HU
- Pooled osteopenia threshold, high
- 166.7 HU
- Pooled osteopenia threshold, low
- 111.2 HU
- Pooled osteoporosis threshold, balanced
- 104.6 HU
- Pooled osteoporosis threshold, high
- 122.3 HU
- Pooled osteoporosis threshold, low
- 86.8 HU
Further pooled results
| Outcome | SMD | P value |
|---|---|---|
| HU values, low BMD vs normal and osteoporosis vs non-osteoporotic boneLow BMD vs Normal | −2.78 | P < 0.001 |
| HU values, low BMD vs normal and osteoporosis vs non-osteoporotic boneOsteoporosis vs Non-osteoporotic bone | −1.57 | P < 0.001 |
Relative risk for low bone mineral density (BMD) by HU
- Decreasing HU associated with stepwise increases in relative risk for low BMD
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Diagnostic performance of lumbar computed tomography Hounsfield unit thresholds for osteoporosis and osteopenia: a systematic review and meta-analysis. Osteoporos Int. Published 10 April 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →