Page 136 - The diagnostic work-up of women with postmenopausal bleeding

Chapter 7
For studies that had numbers available, we constructed 2 x 2 tables and calculated sensitivity and specificity for the diagnosis of endometrial cancer, atypical hyperplasia or endometrial disease. Sensitivity and specificity were calculated for the cases in which both endometrial sampling as well as the reference test was successful.We calculated the weighted sensitivity, taking into account the size of each study, compared to the two different reference strategies.When a 2x2 table could be constructed, we plotted the sensitivity against the ‘1- specificity’ in a receiver-operating curve (ROC).
Results
Study selection
Our systematic search identified 499 titles. After exclusion of studies, which did not exist online anymore and exclusion of duplicates, we identified 377 ar ticles, of which 65 articles were found to be relevant (Figure 1).After reading these 65 articles in full-text, we could include 11 studies that reported on postmenopausal women only, 2 articles that described data on postmenopausal women separately in a total population of perimenopausal women and 17 articles that compared the results of endometrial sampling with histology findings from (blind) D&C or diagnostic hysteroscopy in a combined population of pre- and postmenopausal women.
In none of the 17 studies that repor ted on a combination of pre- and postmenopausal women, we were able to contact the corresponding author. In 10 studies these authors did not respond, while contact details were not available for the other 7 studies.Therefore, we had to exclude these 17 studies from the meta-analysis.The initial agreement of the two reviewers (NvH and MMP) regarding eligibility was 94% (weighted kappa 0.88 (95% CI 0.76-0.99)).
134