Automated CT Screening Predicted Fractures and Mortality in Older Inpatients

This multicentre retrospective cohort study included 46,219 patients aged 60 years or older who had an inpatient CT scan covering the lumbar spine for any indication at hospitals in the Paris region. Fully automated software measured an L1 bone score and identified prevalent lumbar vertebral fractures. Researchers linked hospital data with French health insurance records and assessed osteoporotic fractures requiring hospitalization and mortality over three years.

The cohort’s mean age was 75.3 years, and 27.3% had a lumbar vertebral fracture detected on the index scan. Over three years, 6.8% experienced a fracture requiring hospitalization and 27.9% died. Models using age, sex, bone score and vertebral-fracture detection had areas under the ROC curve of 0.68 for any such fracture and 0.71 for hip fracture. A detected vertebral fracture was associated with a higher incidence of any fracture requiring hospitalization (sub-hazard ratio 2.22, 95% CI 2.06–2.39) and with higher mortality (adjusted hazard ratio 1.29, 95% CI 1.25–1.34).

The findings suggest that routinely acquired CT scans may contain information relevant to fracture-risk identification. However, the observational study assessed prediction, not whether screening or subsequent treatment improves outcomes. Its population was limited to inpatients with eligible scans at one hospital network, and the reported discrimination for fractures was moderate.