Evaluation of screening accuracy on osteoporosis self-assessment tool for Asians and its cut-off value in healthy physical examination population

Abstract

To explore the screening value of osteoporosis self-assessment tool for Asians (OSTA) and the optimal cut-off value in Chinese healthy physical examination population. We selected a healthy physical examination population for bone mineral density screening at the Health Examination Center in Peking University Third Hospital from 2013 to 2016. Quantitative ultrasound (QUS) results were used as the gold standard, and T value ≤-2.5 was defined as osteoporosis patients. Diagnostic test methods were used to analyze the sensitivity, specificity, likelihood ratio and area under curve (AUC) of different cut points of OSTA. The screening accuracy of OSTA at different cut points was compared and the optimal cut-point value determined. A total of 5 833 subjects were included in the study, with an average age of (48.3±17.5) years and 2 594 women (44.5%). The QUS test showed 403 patients with osteoporosis (6.9% of the total population), 343 female osteoporosis patients (13.22% of the female population). In the whole age group, AUC at the international routine cut-off value (OSTA ≤-1) screening for osteoporosis was 0.815 (95%CI: 0.804-0.825), and screening accuracy was higher in the women (AUC=0.837, 95%CI: 0.823-0.851) than that in the men (AUC=0.767, 95%CI: 0.752-0.781; P<0.05). In the whole age group, when the optimal cut-off value was 0, its AUC 0.842 (95%CI: 0.832-0.851) was significantly higher than that when the cut-off value was -1 (P<0.01), and net reclassification improvement (NRI) increased by 5.5%. In the 40 to 65-year-old group, when OSTA cut-off value ≤0, the screening accuracy was significantly higher (NRI=19.5%, P=0.003) than that when it was -1. The OSTA screening tool had good osteoporosis screening value in healthy people, and the screening accuracy in women is higher than that in men. Increasing the screening cut-off value of OSTA would be helpful to improve the screening accuracy in the whole and 40 to 65-year-old population. There may be different optimal cut-off values for different age group population.