Page 143 - The diagnostic work-up of women with postmenopausal bleeding

C1. ROC plot for endometrial disease with D&C as a reference 12:27 Friday, August 21, 2015 1
SROC based on Bivariate model (data for SR van Hanegem Pipelle DC for ED - succesful samples)
C2. ROC plot for endometrial disease with hysteroscopy as a reference 12:27 Friday, August 21, 2015 1
SROC based on Bivariate model (data for SR van Hanegem Pipelle Hyst for ED - succesful samples)
The accuracy of endometrial sampling
1.0 0.8 0.6 0.4 0.2
0.0
0.0 0.2
0.4 0.6
1-specificity Analyzed August 21, 2015
0.8 1.0
1.0 0.8 0.6 0.4 0.2
0.0
0.0 0.2
0.4 0.6
1-specificity Analyzed August 21, 2015
0.8 1.0
Discussion
In this meta-analysis, we assessed the diagnostic accuracy of endometrial sampling
regarding the diagnoses of endometrial cancer, endometrial (pre) cancer and
endometrial disease (including endometrial polyps) in women with PMB,compared to
two different reference strategies: D&C and hysteroscopy. Specificity of endometrial
sampling is very high, irrespective of the type of disease or the reference test that
was used. Sensitivities, on the other hand, are lower than anticipated based on 7 existing meta-analyses, for all types of disease, but especially for atypical hyperplasia
and endometrial disease, which includes endometrial polyps.
An important strength of this meta-analysis is that we performed a thorough
search for articles on the diagnostic accuracy in women with PMB. By searching with all synonyms for PMB and endometrial sampling, we think we selected all articles on this subject.We also selected articles, which described only a subgroup of postmenopausal women and tried to contact the authors of these articles. Unfortunately, none of them responded.We included all eligible articles, regardless of the language used.
This article also has several limitations. Publication bias and the risk of missing potentially relevant articles are concerns with any systematic review.We attempted to mitigate this issue by using a robust search strategy, by checking cross-references and by consulting with a clinical librarian.Also, observer agreement regarding study selection was high. However, by performing this rigorous systematic search, we
141
Sensitivity
Sensitivity