Glenoid neck Hounsfield units on computed tomography can accurately identify patients with low bone mineral density

2018 
Background Osteoporosis is a costly and morbid disease with the first presentation often with a fragility fracture. The purpose of this study was to assess whether Hounsfield unit (HU) measurements on shoulder computed tomography could identify patients at risk of osteoporosis and aid in its diagnosis. Methods We identified patients who had both a computed tomography scan of the glenoid and a dual-energy x-ray absorptiometry scan. Dual-energy x-ray absorptiometry results and HU measurements of the patients' glenoid were recorded. Differences in HU measurements between patients with normal and abnormal central bone mineral density (BMD) were assessed. Correlations were calculated, and receiver operating characteristics were examined. Results A total of 51 glenoids met the criteria. The mean glenoid HU measurement was 140.6 (95% confidence interval [CI], 120.1-161.1) in the osteoporotic group, 168.1 (95% CI, 152.7-183.5) in the osteopenic group, and 233.2 (95% CI, 210.1-256.4) in the normal BMD group ( P t scores in the femoral neck ( r  = 0.581), total hip ( r  = 0.524), and lumbar spine ( r  = 0.345). The area under the receiver operating characteristic curve was 0.918. With 197 HUs used as the cutoff for diagnosis of abnormal BMD, the positive predictive value was 96.6%. With 257.1 HUs used as the cutoff, the negative predictive value was 100%. Conclusion A patient with an HU measurement below 197 has a 97% chance of having low BMD, and a patient with a measurement over 257 likely has normal BMD. In patients with measurements between these values, a definitive diagnosis should be aggressively pursued. Opportunistic screening for a modifiable disease that has significant morbidity and mortality rates at no additional cost, radiation, or time is of great value.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    27
    References
    7
    Citations
    NaN
    KQI
    []