Background: DXA scanner derived images of the spine are used for vertebral fracture detection in adults. It is unknown whether the method could be used in pediatrics. This study evaluated the diagnostic accuracy of DXA images in vertebral fracture assessment (VFA) in children.
Methods: The study included 65 children (37 males; median age 12.1 years) with primary or secondary osteoporosis. Data on clinical history were collected from hospital records. Patients were assessed for spinal compression fractures by standard spinal radiographs and by bone densitometry (Hologic Discovery A) derived VFA images. The visibility and morphology of each vertebra in VFA images was assessed by two readers and by a semi-computerized software developed for the DXA scanner. The findings were compared with those in spinal radiographs and correlated with clinical parameters.
Results: The visibility of vertebrae in VFA images was good in T8-L4 but compromised in the upper thoracic region (T4-T7) and was constantly inferior to that in standard radiographs. A total of 25 vertebral fractures were diagnosed in radiographs, but only 9 (36%) of these also in VFA images. The semi-computerized software could not accurately detect vertebrae in most of the children; accuracy increased with increasing age, height and BMD but was not sufficient to detect vertebral fractures.
Conclusions: The utility of DXA scanner derived images of the spine in vertebral fracture detection in children is limited by compromised visibility and poor diagnostic accuracy. The semi-computerized software is not suitable for pediatric use. These limitations should be kept in mind when assessing pediatric patients for osteoporosis.