Comparing the performance of two clinical models in estimating the risk of endometrial cancer in symptomatic postmenopausal women

Musonda, Patrick, Burbos, Nikolaos, Duncan, Timothy J., Crocker, Simon G., Morris, Edward P. and Nieto, Joaquin J. (2011) Comparing the performance of two clinical models in estimating the risk of endometrial cancer in symptomatic postmenopausal women. European Journal of Obstetrics & Gynecology and Reproductive Biology, 159 (2). pp. 433-438.

Full text not available from this repository.

Abstract

Objective The aim of this study was to internally evaluate the accuracy measures of the two newly developed predictive models, called DEFAB and DFAB, used to estimate the risk of endometrial cancer in postmenopausal women presenting with vaginal bleeding. Study design Prospective study including postmenopausal women presenting with vaginal bleeding. Results Over a 46-month-period, 3795 postmenopausal women presented with vaginal bleeding and were included in the study. A total of 221 (6%) women were diagnosed with endometrial carcinoma. The DEFAB predictive model incorporates known risk factors such as presence of Diabetes, Endometrial thickness measurement on transvaginal ultrasonography, Frequency of bleeding, Age, and Body mass index. The DFAB model is based on the above clinical characteristics excluding the ultrasonography result. For the recommended cut-off values, there was no evidence (p-value = 0.221) of a difference in the diagnostic ability with respect to sensitivity, specificity, area under receiver operating curve, positive predictive value and negative predictive value. There was strong evidence (p-value < 0.0001) to suggest that the diagnostic ability of DEFAB and DFAB agree as evidenced by the excellent Kappa statistic 0.950 (95% CI 0.940–0.960). We found strong evidence (p-value < 0.0001) that the variables incorporated in both predictive models simultaneously correctly classify an individual to either having cancer or not having cancer with respect to logistic discriminant analysis. Conclusion We recommend that these two predictive models can be used interchangeably.

Item Type: Article
Faculty \ School: Faculty of Medicine and Health Sciences > Norwich Medical School
Depositing User: Rhiannon Harvey
Date Deposited: 22 Feb 2012 16:17
Last Modified: 21 Apr 2020 16:50
URI: https://ueaeprints.uea.ac.uk/id/eprint/37397
DOI: 10.1016/j.ejogrb.2011.09.005

Actions (login required)

View Item View Item