Keshtkar Abbas, Tabatabaie Omidreza, Matin Nassim, Mohammadi Zahra, Ebrahimi Mehdi, Khashayar Patricia, Asadi Mojgan
Osteoporosis Research Center (ORC), Endocrinology and Metabolism Clinical Sciences Research Institute (ECSI), Tehran University of Medical Sciences, Tehran, Iran.
MD-MPH, School of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
Rheumatol Int. 2015 Dec;35(12):1995-2004. doi: 10.1007/s00296-015-3286-1. Epub 2015 May 17.
This study was designed to evaluate seven prescreening osteoporosis models in postmenopausal Iranian women. This study was performed on 8644 postmenopausal women who have been referred for bone mineral densitometry (BMD) in BMD center of Shariati hospital in Tehran between 2001 and 2011. Diagnostic properties of seven prescreening instruments were evaluated. With regard to area under curve (AUC), these models have low accuracy (AUC ≤ 0.65). Considering only femoral neck or total femur area, these models had low accuracy but were more sensitive. Except for three models with sensitivities of ≤65 % in both osteoporosis and fracture threshold, other models were around 70 % sensitive. However, these models were not considered clinically useful regarding their positive predictive values (PPV) especially in BMDs ≤02.5. With regard to different measures of diagnostic property, none of these models were good screening tools for osteoporosis or fracture threshold. Although some of them are sensitive, considering other measures such as PPV indicates that they are not completely useful for clinical use. Attempts should be made for developing newer prescreening methods and calibration of the existing models with regard to the studied population.
本研究旨在评估伊朗绝经后女性的七种骨质疏松症预筛查模型。该研究对2001年至2011年间转诊至德黑兰沙里亚蒂医院骨密度中心进行骨密度测量(BMD)的8644名绝经后女性进行。评估了七种预筛查工具的诊断特性。就曲线下面积(AUC)而言,这些模型的准确性较低(AUC≤0.65)。仅考虑股骨颈或全股骨区域时,这些模型准确性较低,但更具敏感性。除了三种在骨质疏松症和骨折阈值方面敏感性均≤65%的模型外,其他模型的敏感性约为70%。然而,就其阳性预测值(PPV)而言,尤其是在骨密度≤02.5时,这些模型在临床上并无用处。就诊断特性的不同测量指标而言,这些模型均不是骨质疏松症或骨折阈值的良好筛查工具。尽管其中一些模型具有敏感性,但考虑到其他指标如PPV,表明它们在临床应用中并不完全有用。应尝试开发更新的预筛查方法,并针对所研究人群对现有模型进行校准。