Selection of two cut-off points via generalized Youden index and receiving operating characteristic surface to predict preeclampsia using the hemoglobin levels in the first trimester of pregnancy

authors:

avatar Hamid Alavi Majd , avatar Nasrin Borumandnia , * , avatar Ali akbar Khadem Mabudi , avatar Noorosadat Kariman , avatar Nastaran Safavi Ardebili , avatar Abbas Hajifathali

Koomesh: Vol.14, issue 3; 321-326
published online: May 25, 2013
article type: Research Article
received: August 07, 2012
accepted: April 08, 2013

how to cite: Alavi Majd H, Borumandnia N, Khadem Mabudi A A, Kariman N, Safavi Ardebili N, et al. Selection of two cut-off points via generalized Youden index and receiving operating characteristic surface to predict preeclampsia using the hemoglobin levels in the first trimester of pregnancy. koomesh. 2013;14(3):e152583. 

Abstract

 Introduction: Researchers are extensively using biomarkers to detect diseases. Commonly, a cut-off point is selected for separating sick and healthy people. Sometimes it is necessary to obtain more than one cut-off points, by classifying subjects into more than two groups. As such, in the present paper, generalized Youden index has been used to classify subjects into three groups. The main intention of this study was to determine the optimal cut-off points for the hemoglobin levels in the first trimester of pregnancy in order to predict preeclampsia. Materials and Methods: We used data of the hemoglobin levels in the first trimester of pregnancy and preeclampsia from 620 pregnant women who were referred to Tehran's Milad Hospital in 2009-2010. The optimal cut-off points for prediction of preeclampsia in the first trimester were obtained by using generalized Youden index and volume under the receiving operating characteristics (ROC) surface (VUS). Statistical analysis was performed using by R software version 2-15-1 and DiagTest3Grp package.  Results: The estimated cut-off points, using by Youden index, were 13.18 and 14.5 with correct classification proportions of 70%, 35% and 15%, respectively. The estimated cut-off points, using by ROC surface were 12.4 and 13.1 and correct classification proportions were 42%, 35% and 55%, in order. The volume under the ROC surface was 0.25. Conclusion: The generalized Youden index, as a complementary index for VUS, can be adopted to achieve greeter diagnostic accuracy it can also be used to achieve a three group classification of subjects