POS1117 AGREEMENT BETWEEN VERTEBRAL FRACTURE ASSESSMENT AND RADIOGRAPHIC SCANS FOR THE DIAGNOSIS OF VERTEBRAL FRACTURE

Abstract

Background:Conventional radiography of thoracic and lumbar spine are considered the gold-Standard imaging for vertebral fracture (VF) identification. However, there was a growing interest in the use of vertebral fracture assessment (VFA) as a low radiation tool. In fact, the radiation dose for VFA is about 3 micro Sievert compared to 600 Sievert for lateral standard radiography. Moreover, it costs 2-times less than X-rays. However, the advantage granted by a lower radiation dose is unfortunately counterbalanced by higher noise rates and therefore lower image quality.Objectives:The aim of this study was to investigate the diagnostic accuracy of VFA compared to lateral spine radiographs.Methods:We conducted a cross-sectional study over a period of 5 months at the rheumatology department. The study included post-menopausal women without a previous diagnosis of VF. Each participant had a BMD assessment, a VFA scan and a lateral thoracolumbar X-rays (Rx) to detect VF. VF were identified according the Genant semi-quantitative method. The number of unreadable vertebrae were compared between VFA and Rx using a McNemar test. Cohen’s kappa coefficient, the sensitivity (Se), the specificity (Sp), the positive (PPV) and negative predictive (NPV)values at each vertebral level and at the individual level were calculated with 95% confidence interval (95% CI) to analyze the agreement between VFA and Rx results.Results:The study included 62 patients were collected. The mean age was 62.03±7.84 years. the mean body mass index (BMI) was 29.99±5.13 kg/m2 and the mean menopausal duration was 15.7±8.28 years. Parental history of hip fracture and prior history of fragility fracture were recorded in 25.9% and 40.3% respectively. A premature menopause, Rheumatoid arthritis and use of corticosteroids were found in 12.9%, 4.8% and 19.4% respectively. A historical height loss of more than 4cm and a prospective height loss of more than 2 cm were reported in 21% and 38.7% respectively. Using VFA, 22.6% of our population had at least one VF≥ grade2 and 33.9% had at least one VF≥grade1. Using Rx, 19.4% of our population had at least one VF≥ grade2 and 25.8% had at least one VF≥grade1. Among all vertebral levels, no statistically significant difference was found while comparing the number of unreadable vertebrae on VFA and radiographs. Taking into consideration VF≥grade2, there was an almost perfect agreement between VFA and Rx (k=0.806, p<0.001); the Se, Sp, PPV and NPV of VFA were respectively 91.7% (95% CI 59.8-99.6), 94%(95% CI 82.5-98.4), 78.6% (95% CI 48.8-94.3) and 97.9%(95% CI 87.5-99.9). . Taking into consideration VF≥grade1, there was a moderate agreement between VFA and Rx (k=0580, p<0.001); the Se, Sp,PPV and NPV of VFA were respectively 81.3% (95% CI 53.7-95),82.6% (95% CI 68-91.7), 61.9% (95% CI 38.7-81) and 92.7% (95% CI 79-98.1). At the vertebral level and when including only grade 2 and 3 VF, the Se, Sp, PPV of VFA were 100% (95% CI 82.2-100), 98.9% (95% CI 97.7-99.5), 74.2% (95% CI 55.1-87.5) and 100% (95% CI 99.3-100) respectively. At the vertebral level, the Se, Sp, PPV and NPV of VFA in detection of grade≥1 VF were 65.8% (95% CI 48.6-79.9), 98.3% (95% CI 96.9-99.1), 67.6% (95% CI 50.1-81.4) and 98.1% (95% CI 96.8-99.0) respectively. The analysis by vertebral level revealed substantial to almost perfect agreement for all levels except for T4 and T6 where the test was not applicable. There was a perfect agreement for T5, T8, L3 and L4 vertebrae.Conclusion:VFA showed a high diagnostic performance compared to Rx. One interesting finding is that the NPV of VFA is very high at the vertebral and the individual level and either grade 1 VF were took into consideration or not. This means that a normal VFA can formally rule out the presence of VF.Disclosure of Interests:None declared.