Misdiagnosis of vertebral fractures on local radiographic readings of the multicentre POINT (Prevalence of Osteoporosis in INTernal medicine) study

Abstract

Background: Osteoporotic vertebral fractures (VFs) are often misdiagnosed because asymptomatic and occurring in the absence of specific trauma. Further, diagnostic assessment of VFs may be suboptimal.

Aim of the study: To assess the misdiagnosis of vertebral fractures on local radiographic readings in the cohort of patients enrolled in the POINT study.

Methods: We enrolled hospitalised patients, admitted for any cause to the Internal Medicine Units of 37 hospitals participating to the cross-sectional previously published POINT study. The assessment of VFs was performed both by local radiologists and by two expert skeletal radiologists, by using semiquantitative method (SQ). To better evaluate mild vertebral deformities, the two central radiologists also used the algorithm-based qualitative assessment (ABQ).

Results: The radiographs of 661 patients (401 females; mean age 75.8±8.0) were evaluated. The inter-reader percent agreement between two central expert radiologists per-vertebra assessment was excellent (99.78%; k=0.984; 95% CI, 0.977-0.991). Central reading identified 318/661 (48.1%) patients with at least one VF. Local and central readings agreed in 502/661 (75.9%) patients, resulting in a fair reproducibility (k=0.52; 95% confidence interval 0.44-0.59). Diagnostic performance parameters of local readings were: sensitivity 76.1%; specificity 75.8%; PPV 74.46%; NPV 77.38%). By examining 9254 vertebrae, central and local readers diagnosed 665 (7.2%) and 562 (6.1%) VFs respectively. Misdiagnosis (102 false positives and 205 false negatives) mainly occurred for mild VFs. Local readings identified correctly 460 out 665 VFs diagnosed by central readings, resulting in sensitivity of 69.2% and PPV of 81.8%.

Conclusions: Following a standardized protocol of acquisition techniques and of interpretation criteria, an excellent agreement between local and central readings for moderate and severe vertebral fractures resulted. However a significant amount of mild vertebral fractures, that are the most of VFs, were misdiagnosed by local radiologists. In order to improve VFs assessment, the radiologists should be trained and sensitized in relation to the relevant clinical significance of osteoporotic VFs identification.

Keywords: Algorithm-based qualitative approach (ABQ); Osteoporosis; Semiquantitative method (SQ); Vertebral deformities; Vertebral fractures.