Diagnostic accuracy of DXA compared to conventional spine radiographs for the detection of vertebral fractures in children.

OBJECTIVES: In children, radiography is performed to diagnose vertebral fractures and dual energy x-ray absorptiometry (DXA) to assess bone density. In adults, DXA assesses both. We aimed to establish whether DXA can replace spine radiographs in assessment of paediatric vertebral fractures. METHODS: Prospectively, lateral spine radiographs and lateral spine DXA of 250 children performed on the same day were independently scored by three radiologists using the simplified algorithm-based qualitative technique and blinded to results of the other modality. Consensus radiograph read and second read of 100 random images were performed. Diagnostic accuracy, inter/intraobserver and intermodality agreements, patient/carer experience and radiation dose were assessed. RESULTS: Average sensitivity and specificity (95 % confidence interval) in diagnosing one or more vertebral fractures requiring treatment was 70 % (58-82 %) and 97 % (94-100 %) respectively for DXA and 74 % (55-93 %) and 96 % (95-98 %) for radiographs. Fleiss' kappa for interobserver and average kappa for intraobserver reliability were 0.371 and 0.631 respectively for DXA and 0.418 and 0.621 for radiographs. Average effective dose was 41.9 µSv for DXA and 232.7 µSv for radiographs. Image quality was similar. CONCLUSION: Given comparable image quality and non-inferior diagnostic accuracy, lateral spine DXA should replace conventional radiographs for assessment of vertebral fractures in children. KEY POINTS: • Vertebral fracture diagnostic accuracy of lateral spine DXA is non-inferior to radiographs. • The rate of unreadable vertebrae for DXA is lower than for radiographs. • Effective dose of DXA is significantly lower than radiographs. • Children prefer DXA to radiographs. • Given the above, DXA should replace radiographs for paediatric vertebral fracture assessment.

Diagnosis of vertebral fractures in children: is a simplified algorithm-based qualitative technique reliable?

BACKGROUND: Identification of osteoporotic vertebral fractures allows treatment opportunity reducing future risk. There is no agreed standardised method for diagnosing paediatric vertebral fractures. OBJECTIVE: To evaluate the precision of a modified adult algorithm-based qualitative (ABQ) technique, applicable to children with primary or secondary osteoporosis. MATERIALS AND METHODS: Three radiologists independently assessed lateral spine radiographs of 50 children with suspected reduction in bone mineral density using a modified ABQ scoring system and following simplification to include only clinically relevant parameters, a simplified ABQ score. A final consensus of all observers using simplified ABQ was performed as a reference standard for fracture characterisation. Kappa was calculated for interobserver agreement of the components of both scoring systems and intraobserver agreement of simplified ABQ based on a second read of 29 randomly selected images. RESULTS: Interobserver Kappa for modified ABQ scoring for fracture detection, severity and shape ranged from 0.34 to 0.49 Kappa for abnormal endplate and position assessment was 0.27 to 0.38. Inter- and intraobserver Kappa for simplified ABQ scoring for fracture detection and grade ranged from 0.37 to 0.46 and 0.45 to 0.56, respectively. Inter- and intraobserver Kappa for affected endplate ranged from 0.31 to 0.41 and 0.45 to 0.51, respectively. Subjectively, observers' felt simplified ABQ was easier and less time-consuming. CONCLUSION: Observer reliability of modified and simplified ABQ was similar, with slight to moderate agreement for fracture detection and grade/severity. Due to subjective preference for simplified ABQ, we suggest its use as a semi-objective measure of diagnosing paediatric vertebral fractures.

Comparison of clinically suspected injuries with injuries detected at whole-body CT in suspected multi-trauma victims.

AIM: To assess the accuracy of the trauma team leader's clinical suspicion of injury in patients who have undergone whole-body computed tomography (WBCT) for suspected polytrauma, and to assess the frequency of unsuspected injuries and specific patterns of injury at WBCT. MATERIALS AND METHODS: Requests for patients who underwent WBCT for suspected polytrauma from April 2011 to March 2014 were reviewed and each body area that was clinically suspected to be injured was recorded. Body areas were divided into the following groups based on anatomical regions covered in each segment of the WBCT examination: head (including facial bones); neck (including cervical spine); chest (including thoracic spine); and abdomen/pelvis (including lumbar spine). The formal radiology report for each study was reviewed and injuries found at CT were grouped into the same body areas. For each patient, the number of clinically suspected injured areas was compared to the number of confirmed injured areas at WBCT. RESULTS: Five hundred and eighty-eight patients were included in the study. Thirty-two percent (186/588) had a normal scan. Ninety-three percent (546/588) had fewer injured body areas at WBCT than suspected. Four percent (27/588) had the equivalent number of injured areas at WBCT as suspected. Three percent (15/588) had more injured areas at WBCT than suspected. Fifty percent (263/527) with clinically suspected chest injuries were confirmed to have chest injuries at WBCT. This was lower for other areas: abdomen/pelvis 31% (165/535); head 29% (155/533); neck 13% (66/513). Four percent of (24/588) patients had unsuspected injuries found at WBCT. Seventy-five percent (18/24) of unsuspected injuries were considered as serious, where failure to treat would have the potential for significant morbidity. Most of these patients had severe injuries to other body areas that were correctly suspected. Of the 165 with abdominal/pelvic region injuries, there were associated injuries in the thoracic region in 62% (103/165) cases. Of the 278 thoracic injuries, there were abdominal/pelvic region injuries in 37% (103/278) and injuries to the neck in 19% (52/278) cases. Of the 158 head injuries, there were neck injuries in 22% (34/158) cases. CONCLUSION: Clinical suspicion of injury correlates poorly with findings at WBCT, with a high proportion of uninjured body areas. The number of unsuspected injuries found at WBCT was low, but the majority of these were serious injuries, possibly masked by distracting injury to other body areas. The use of a WBCT protocol is recommended for suspected polytrauma, but regular monitoring of WBCT findings and regular feedback of the results to emergency physicians is suggested to help inform their selection of patients for trauma WBCT.