Page 138 - The diagnostic work-up of women with postmenopausal bleeding

Chapter 7
Table 1. Study characteristics of the twelve included studies on the diagnostic accuracy of endometrial sampling
Study design
Patient selection
Index test
Reference standard
Menopausal status
PMB n
HRT %
Goldberg 1982
Batool 1994
Ben-Baruch 1994
vdBosch 1995
vdBosch 1996
Giusa-Chifieri 1996
Gupta 1996
De Silva 1997
Mortakis 1997
Bunyavejchevin 2001
Epstein 2001 Spicer 2006
Prospective
Prospective Retrospective
Prospective
Prospective
Prospective
Prospective
Prospective
Not reported
Prospective
Prospective Prospective
Unclear
Unclear Unclear
Consecutive
Consecutive
Unclear
Consecutive
Consecutive
Unclear
Unclear
Consecutive Consecutive
Vabra & Accurette
Pipelle Pipelle
Pipelle
Pipelle
Novak
Pipelle
Pipelle
Pipelle
Pipelle
Endorette Accurette
Blind D&C
Blind D&C Blind D&C
Hysteroscopy w/ histology
Hysteroscopy w/ histology
Hysteroscopy w/ histology
Hysteroscopy w/ histology
Hysteroscopy w/ histology
Hysteroscopy w/ histology
Blind D&C
Blind D&C
Hysteroscopy w/ histology
Post
Post
Pre & post
Post
Post
Post
Post
Post
Pre & post Post
Post Post
40 Nr
70 Nr 90 Nr
140 0
87 0
80 Nr
76 0
50 Nr
78 0
30 0
133 56 136 Nr
D&C=dilatation and curettage; PMB=number of women with postmenopausal bleeding; Nr=not reported
Quality assessment
Table 2 presents quality assessment of the included studies.12-24 Quality assessment showed in four studies (25%) a ‘low’ risk of bias on all four items, three studies showed a ‘high’ risk of bias on one of the items, while eight studies had an ‘unclear’ risk of bias on the description of methods on patient selection, the index or reference test. All studies, except for one scored ‘low’ on the three items of applicability.12-23
Based on a high concern of applicability of the index test and reference standard described in O’Connell et al, we decided to exclude this study.24 After study selection and quality assessment, we included 12 articles in this systematic review, reporting on 1,029 women with postmenopausal bleeding (Table 1 and 2, Figure 2).
136