T-score differences and nonprogression in lumbar vertebrae as predictors of vertebral fractures

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review


Purpose: In case of a vertebral fracture, the area of the vertebrae decreases with a concomitant increase in BMD, as assessed by a DXA scanning. Furthermore, a vertebral fracture may disrupt the normal increase in vertebral body area from L1 to L4 (nonprogression). We aimed to examine associations between T-score difference and nonprogression of vertebral area and vertebral fractures. Methods: We identified 100 patients with 1 or more fractures in L1-L4 and 106 patients without fractures. All patients had undergone a DXA scan and a lumbar spine X-ray. In fracture patients, we recorded T-score difference between the fractured vertebra and the adjacent vertebra, and whether the fractured vertebra was smaller than the one above (nonprogression). In nonfracture patients, the greatest positive T-score difference was recorded, and nonprogression was present if vertebral area did not increase successively from L1 to L4. Results: With a T-score difference ≥1 SD odds ratio for fracture was 1.30 (0.74-2.29). Sensitivity and specificity were 0.40 and 0.66, respectively. With T-score difference ≥1.5 SD, odds ratio for fracture was 2.26 (1.08-4.73). Sensitivity and specificity were 0.24 and 0.88, respectively. Nonprogression was very common in the no-fracture group (38%), while only 23% of X-ray verified fractures had nonprogression. Conclusion: A randomly found T-score difference ≥1.5 SD between adjacent vertebrae on a DXA scan is associated with a small increased risk of compression fracture. Nonprogression is very common in patients without fractures.

Original languageEnglish
JournalClinical Endocrinology
Pages (from-to)58-62
Number of pages5
Publication statusPublished - Jul 2019

    Research areas

  • DXA scan, nonprogression, predictors, T-score, vertebral fracture

See relations at Aarhus University Citationformats

ID: 157539511