Prevalence and predictors of silent vertebral compression fractures using DXA: a UK Biobank Imaging Study - Summary - MDSpire
Coming Soon: Introducing MDSpire News. Learn more
Conexiant’s news site is now MDSpire News. Learn more

Prevalence and Predictors of Asymptomatic Vertebral Compression Fractures on DXA: Findings from UK Biobank Imaging

  • By

  • David John Wilson

  • Richard William Whitehouse

  • Georgina Marian Allen

  • Fiona Saunders

  • Iain Wilson

  • Richard Abel

  • October 13, 2026

Share

Objective:

Estimate the prevalence of silent vertebral fractures and assess demographic and clinical predictors among UK Biobank adults aged 40–79 years without reported spinal fracture or trauma.

Approach:
  • Study population: Whole-spine DXA images and participant data were analysed for 2,446 eligible adults (mean age 62.0 years; 1,151 men).
  • Fracture assessment: An automated six-point morphometric algorithm was reviewed by a trained radiographer and independently verified by a musculoskeletal radiologist blinded to clinical information. Fractures with at least 25% vertebral height loss were classified as moderate or s…
  • Analysis: The study assessed prevalence and associations with participant characteristics using multivariable logistic regression; predictive performance was also examined with logistic regression and random forest models.
Key Findings:
  • Moderate or severe vertebral fractures were identified in 356 participants (14.6%).
  • Prevalence increased from 9.4% among participants aged 40–49 years to 19.6% among those aged 70–79 years.
  • Prevalence was higher in men than women (18.8% versus 10.8%).
  • Radiologist adjudication classified fractures as mild in 42.3%, moderate in 53.5%, and severe in 4.2%; including 20–24.9% deformities yielded an approximately 30% prevalence.
  • Interobserver agreement between radiologists was substantial (Fleiss’ κ = 0.74), while agreement between radiologist adjudication and the automated system was lower (κ = 0.31).
Interpretation:

In this UK Biobank imaging cohort, silent vertebral fractures were identified in 14.6% of participants using a threshold of at least 25% height loss. The authors report that degenerative changes and spinal shape variation contributed to automated false-positive classifications.

Limitations:
  • The analysis included 2,446 participants from the UK Biobank imaging cohort, rather than the full UK Biobank population.
  • The supplied context does not report the adjusted predictor estimates or the predictive-model performance results.
  • DXA image quality was inadequate for 220 of 32,058 vertebrae assessed.
Sources:

Original Source(s)

Related Content