Page 72 - The diagnostic work-up of women with postmenopausal bleeding

Chapter 4
Abstract
Objective To externally validate two previously developed prediction models that estimate the probability of endometrial cancer in women with postmenopausal bleeding.
Design and setting We performed an external validation study of two previously developed prediction models in two independent datasets of consecutive women not using hormone replacement therapy with a first episode of postmenopausal bleeding.
Population We studied women with postmenopausal bleeding. One dataset (559 women) was prospectively collected in three general hospitals in the Netherlands including, the other dataset (433 women) was prospectively collected in a university hospital in Sweden.
Methods We retrospectively evaluated two models that predict endometrial cancer in the two validation databases. We then evaluated three diagnostic strategies, a ‘patient characteristics’ rule, based on characteristics of the women without transvaginal ultrasound, a ‘sequential’ rule, i.e. ultrasound in case the probability for cancer exceeded 4% based on characteristics, and subsequent histological analyses when the endometrial thickness exceeded 4 mm and an ‘integrated’ rule with a probability estimate based on both characteristics of the women and ultrasound results and endometrial sampling when the probability of cancer exceeded 4%.
Main outcome measures We studied the performance of the models in terms of discrimination and calibration. We then calculated the number of carcinomas detected and missed using the three different strategies, as well as the number of ultrasounds and invasive procedures performed with the three different strategies.
Results In both the Dutch and the Swedish databases, the two models showed good performance in terms of discrimination and calibration.The three strategies, based on these two models, all detected all women with endometrial (pre) cancer. Applying the ‘integrated’ or ‘sequential’ strategy would, compared to current practice (ultrasound only), leads to a 3 to 6 % decrease in the number of women in need for further invasive testing.
70