Prevalence and predictors of silent vertebral compression fractures using DXA: a UK Biobank Imaging Study

OBJECTIVES: Silent vertebral fractures are common but under-recognised. We aimed to estimate the prevalence in an asymptomatic population and to identify demographic and clinical predictors using the UK Biobank data. MATERIALS AND METHODS: DXA spine images from 2446 asymptomatic UK Biobank participants aged 40-79 years were analysed retrospectively. Vertebral fractures were defined as moderate or severe deformities (≥ 25% vertebral height loss). Fractures were identified using an artificial-intelligence-assisted six-point morphometric algorithm with manual correction and radiologist confirmation. Associations with age, sex, bone mineral density (BMD), body mass index (BMI), medication, and back pain were analysed. RESULTS: A total of 2446 participants (mean age 62.0 ± 7.4 years; 1151 men) were analysed. Moderate or severe vertebral fractures were identified in 356 participants (14.6%). Prevalence increased with age, from 9.4% in participants aged 40-49 years to 19.6% in those aged 70-79 years. Fractures were more common in men than women (18.8% vs. 10.8%; adjusted OR 0.51 for women vs. men, 95% CI 0.40-0.66; p < 0.001). BMD and most conventional clinical risk factors had limited predictive value. CONCLUSIONS: Silent vertebral fractures are common in midlife, frequently occur in individuals with normal bone density, and are more prevalent in men. Conventional clinical risk factors have a limited ability to identify affected individuals. KEY POINTS: Question Silent vertebral fractures are frequently overlooked in asymptomatic individuals, and their prevalence and demographic distribution in middle-aged adults are not well-characterised. Findings In a large asymptomatic cohort, vertebral fractures were common, showed male predominance, and were frequently present in individuals without low bone mineral density or recognised clinical risk factors. Clinical relevance Silent vertebral fractures are prevalent in asymptomatic adults and are poorly captured by conventional risk assessment, indicating limitations of current approaches to fracture identification.