Table 4.4Tests for exclusion of underlying conditions – comparative studies

Bibliographic informationStudy type and evidence levelNo. of patientsPatient characteristicsIntervention and comparisonFollow-upOutcome measuresEffect sizeSource of funding and additional comments
Baxter 2002156randomised; single-blind; concealed

EL = Ib
96 asked to enter study, 83 entered study, 40 in flexible and 43 in rigid group.Population characteristics: Women; referred for hysteroscopy due to AUB; > 16 years old; not pregnant.

Baseline characteristics (flexible vs rigid):

Age = 49 vs 47
Pre-menopausal = 26 vs 25

Reason for referral:

Menorrhagia = 17 vs 8
Post-menopausal bleeding = 9 vs 10
Post-coital bleeding = 1 vs 1
Intermenstrual bleeding = 0 vs 2
Irregular bleeding = 8 vs 13
Abnormal bleeding on HRT = 5 vs 7
Part of trial = 0 vs 2

Country: UK
Flexible hysteroscopy; rigid hysteroscopy30 minutesPain levelsPain level (10 cm VAS scale):

Immediately after procedure = 1.8 vs 4.0, P = 0.0001

30 minutes after procedure = 1.0 vs 1.7, P = 0.031
Funding source: Not stated
Clark 2002155Systematic review

EL = 2++
3486 studies identified, 208 retrieved for detailed analysis, 65 primary studies included in review.Population characteristics: Search undertaken on Cochrane; MEDLINE; EMBASE, and hand searching of existing reviews.

Country: UK
Hysteroscopy for identifying endometrial cancer and hyperplasia.N/Asensitivity, specificity, PPV, PNVEndometrial cancer: sensitivity = 86.4 (95% CI 84 to 88.4); specificity = 99.2% (95% CI 99.1 to 99.3).

Pre-test prevalence = 3.9%; pre-test likelihood ratio: positive 60.9, negative 0.15. post-test probability: positive = 71.8%; post test negative = 0.6%.

Endometrial disease: sensitivity = 78%; specificity = 95.8%.

Pre-test prevalence = 10.6%; pre-test likelihood ratio: positive = 10.4 and negative = 0.24. post-test probability: positive = 55.2%; post-test negative = 2.8%.

Biopsy or hysterectomy or D&C pathology used as reference standards.

Majority of studies use mixture of post- and pre-menopausal women

Higher quality studies show lower likelihood ratio and outcome probability. All studies (n = 61) show pre-test likelihood ratio of endometrial cancer: positive = 60.9 and negative = 0.15; post-test probability – positive = 71.8, negative = 0.6. However, high quality studies only (n = 11) show pre-test likelihood ratio: positive = 34.8, negative = 0.21, and post-test probability: positive = 58.6, and negative = 0.8 respectively.

For endometrial disease all studies (n = 71) show pre-test likelihood ratio: positive = 10.4 and negative = 0.2; post-test = 55.2, 2.8, and high quality show (n = 12) pre-test likelihood ratio: positive = 5.5, negative = 0.31; post-test probability: positive = 39.4, negative = 3.5.

Factors such as study setting, study population and patient selection impact on results.
Funding source: University of Birmingham Interdisciplinary Research Fund and Birmingham Women's Hospital R&D fund

Study summary: hysteroscopy is good at identifying endometrial cancer, but less so for identifying endometrial disease.
De Kroon 2003154Diagnostic; meta-analysis

EL = Ib
109 studies identified, 24 included in reviewPopulation characteristics: Search strategy: MEDLINE, EMBASE, DARE, Cochrane library, ISI – current contents

Studies examining saline contrast hysterosonography in AUB

Country: Netherlands
Saline contrast hysterosonography for AUBNo follow-upSensitivity, specificity, PPV, PNVPooled results for studies were hysterectomy was the reference method – 2 studies, n = 96 – positive likelihood ratio = 16.8, negative likelihood ratio = 0.05, positive post-test probability = 0.93, negative post-test probability = 0.04

Pooled results for studies were verification bias avoided (quality measure) – 16 studies, n = 877 – sensitivity = 0.95, specificity = 0.88, positive likelihood ratio = 8.23, negative likelihood ratio = 0.06, positive post-test probability = 0.91, negative post-test probability = 0.07

Pooled results for studies for identification fibroidssensitivity = 0.87, specificity = 0.92, positive likelihood ratio = 11.0, negative likelihood ratio = 0.07

Pooled results for studies for identification endometrial polypssensitivity = 0.86, specificity = 0.81, positive likelihood ratio = 5.23, negative likelihood ratio = 0.12

Success rate for SIS in 24 studies = 93%.

5 complications reported in 2278 procedures.
Funding source: Not stated

Study summary: SIS is an accurate method for evaluating uterine cavity in women with AUB.
Dueholm 2002152Systematic review

EL = Ib
18 papersPopulation characteristics: MEDLINE search – 1982 to 2001, English language, diagnostic accuracy of transvaginal ultrasound, hysterosonographic examination, hysteroscopy, magnetic resonance imaging.

Country: Denmark
Accuracy of transvaginal ultrasound, hysterosonographic examination, hysteroscopy, magnetic resonance imaging.No follow-upAccuracy of test – sensitivity, specificityTVS (n = 11 studies) overall diagnostic accuracy – sensitivity = 87% (range 24% to 96%), specificity = 82% (range = 29% to 93%).

For identification of polypssensitivity = 80% (range = 31% to 94%).

For identification of submucous myomas – sensitivity = 94% (range 62% to 100%).

Hysterosonographic examination overall diagnostic accuracy – sensitivity = 94% (range 83% to 100%), specificity = 85% (range = 72% to 99%)

For identification of polypssensitivity = 93% (range 67% to 100%), specificity = 96% (range 93% to 100%).

Hysteroscopy – no combined results reported.

MRI – no combined results shown.

Highlights that many studies use endometrial sampling as gold-standard which is inappropriate.
Funding source: Not stated
Farquhar 2003151systematic review; diagnostic

EL = Ib
19 papersPopulation characteristics: MEDLINE and EMBASE 1980 to 2001. Search terms and MeSH terms used.

Standard inclusion/exclusion criteria used.

Standard quality assessment used.

Standard data extraction used

Country: New Zealand
Transvaginal ultrasound; sonohysterography; hysteroscopy; histopathology – operative hysteroscopy or hysterectomyNo follow-upAccuracy of diagnostic method – sensitivity, specificity, PPV, NPV, +LR, -LR; patient discomfortTransvaginal ultrasound vs histopathology or hysteroscopy for identification of intrauterine pathology (n = 10):
Sensitivity range = 48% to 100%, specificity range = 12% to 100%, LR+ range 1.0 to 51.6, -LR range = 0.05 to 0.79.

Transvaginal ultrasound vs histopathology or hysteroscopy for identification of submucous fibroids:
Sensitivity range = 21% to 100%, specificity range = 53% to 100%, LR+ range 1.6 to 62.3, -LR range = 0.03 to 0.47.

Transvaginal ultrasound vs histopathology or hysteroscopy for identification of hyperplasia:
Sensitivity range = 33% to 100%, specificity range = 79% to 100%, LR+ range 2.6 to 679, -LR range = 0.04 to 1.00.

Sonohysteroscopy vs histopathology or hysteroscopy for identification of intrauterine pathology (n = 11):
Sensitivity range = 85% to 100%, specificity range = 50% to 100%, LR+ range 2.0 to 80.3, -LR range = 0.04 to 0.38.

Sonohysteroscopy vs histopathology or hysteroscopy for identification of submucous fibroids:
Sensitivity range = 57% to 100%, specificity range = 96% to 100%, LR+ range 21.3 to 80.3, -LR range = 0.06 to 0.47.

Sonohysteroscopy vs histopathology or hysteroscopy for identification of endometrial hyperplasia:
Sensitivity range = 29% to 80%, specificity range = 82% to 100%, LR+ range 1.6 to 70.4, -LR range = 0.14 to 29.

Hysteroscopy vs histopathology or hysteroscopy for identification of intrauterine pathology (n = 3):
Sensitivity range = 90% to 97%, specificity range = 62% to 93%, LR+ range 2.55 to 14.56, -LR range = 0.03 to 0.11.

Hysteroscopy vs histopathology or hysteroscopy for identification of submucous fibroids:
Sensitivity range = 53% to 100%, specificity range = 97% to 100%, LR+ range 10.4 to 41.0, -LR range = 0.08 to 0.48.

Hysteroscopy vs histopathology or hysteroscopy for identification of endometrial hyperplasia:
Sensitivity range = 90% to 100%, specificity range = 99% to 100%, LR+ range 47.0 to 111.7, -LR range = 0.02 to 0.15.

Discomfort with transvaginal ultrasound – one study reported 2% of people found it unpleasant, and 40% experienced some discomfort.

Discomfort with sonohysterography – study reported 13% found it unpleasant and 53% had discomfort

Discomfort with hysteroscopy – 1.6% procedures not completed due to intolerance, and 3.6% of people would not have procedure again due to pain.

Safety infrequently reported in studies for any test.
Funding source: Not stated

Study summary: Although high degree of variation in studies, all tests were moderately accurate at identifying pathology.
Guyatt 1992150Systematic review

EL = 2++
55 articles includedPopulation characteristics: MEDLINE

Two searches designed.

Inclusion/exclusion criteria systematically applied.

Systematic quality assessment
Testing for anaemiaN/ALikelihood of anaemia55 studies identified.

Serum ferritin (n = 2579) Area under ROC curve = 0.95 (95% CI 0.94 to 0.96).

Likelihood ratio at serum ferritin levels (μg/l):

> 100 = 0.08
45 to 100 = 0.54
35 to 44 = 1.83
25 to 34 = 2.54
24 to 15 = 8.83
< 15 = 51.85

Red cell protoporphyrin (n = 288) Area under ROC curve = 0.77 (0.71 to 0.83)

Likelihood ratio at red cell protoporphyrin levels (μg/l):

< 50 = 0.12
51 to 150 = 0.56
151 to 250 = 2.01
251 to 350 = 6.05
> 351 = 8.31

Mean cell volume (N = 436) Area under ROC curve =0.76 (0.72 to 0.80)

Likelihood ratio at mean cell volume levels (μm3):

> 90 = 0.29
85 to 89 = 0.76
80 to 84 = 0.91
75 to 79 = 1.00
70 to 74 = 3.33
< 70 = 12.47

Trans-ferrin saturation (n = 764) area under ROC curve = 0.74 (0.70 to 0.78)

Likelihood ratio at transferrin saturation level (%):

> 50 = 0.15
30 to 49 = 0.43
20 to 29 = 0.52
10 to 19 = 0.81
5 to 9 = 2.54
< 5 = 10.46

Red cell volume distribution (n = 273) area under ROC curve = 0.62 (0.55 to 0.69)

Likelihood ratio at red cell volume distribution level:

< 15 = 0.61
15 to 16 = 0.84
17 to 20 = 1.78
21 = 2.72.
Funding source: Not stated

Study summary: Serum ferritin radioimmunoassay is powerful test in identification of iron-deficiency anaemia
James 2004145systematic review; diagnostic studies

EL = 2−
107 articles identifiedPopulation characteristics: MEDLINE. 1990 to 2003. Keyword search only.

Country:
Testing for von Willebrand disease in menorrhagiaN/APrevalence of vWD or platelet abnormalities; sensitivity; specificity5 studies showed prevalence of vWD of 5.3% to 20%. Samples sizes from 19 to 150.

6 studies showed sensitivity of between 79% and 100%

4 studies showed specificity of between 80% to 95%
Funding source: Dade-Behring

Study summary: Inadequate evidence to support routine testing for vWD in menorrhagia
Krassas 199452Epidemiology

EL = 2+
428: 214 with thyroid disease; 214 matched controlsPopulation characteristics: Women; with or without thyroid disease

Country: Greece
Association between thyroid condition and menstrual disordersNo follow-upPresence of thyroid condition – TT3 and TT4 levels; menstrual disorders; smoking status; BMIOf the 214 patients, 168 (78.5%) had regular menstrual cycles and 46 (21.5%) irregular cycles. Out of 214 normal controls, matched for age and weight, 196 (91.6%) had normal menstruation and 18 (8.4%) irregular cycles. 2 (4.5%) and 2 (11%) of thyrotoxic and normal controls had menorrhagia.

No statistical difference between groups.
Funding source: Not stated

Study summary: These data demonstrate that hyperthyroidism in women is less frequently associated with menstrual abnormalities than was previously believed.
Philipp 2003563Cohort; epidemiology

EL = 2+
126: 74 menorrhagia; 52 controlsPopulation characteristics: patient group: women; physician-diagnosed menorrhagia; known pathology excluded; scheduled for hysterectomy during study period; those taking pharmaceutical treatments asked to stop.

Mean age – 40.4 (range 17 to 55)

Controls: women; same as above but no menorrhagia.

Country: USA
platelet functional defects association with menorrhagiaMBLPBAC; platelet function testOf 59 PBACs returned by study group: 51 had score > 100; 37 had score > 185.

Platelet aggregation and ATP release, comparison between study (n = 74) and control (n = 52) groups. Platelet aggregation: epinephrine 16 vs 2 (P = 0.005); ristocetin 20 vs 4 (P = 0.007); collagen 9 vs 2 (P = 0.105); ADP 3 vs 1 (P = 0.5); arachidonic acid 6 vs 1 (P = 0.13). ATP release: ADP 30 vs 7 (P = 0.0009); arachidonic acid 16 vs 1 (P = 0.001); collagen 18 vs 5 (P = 0.04); thrombin 1 vs 0 (N/A).
Funding source: Association of Teachers Preventative Medicine grant

Study summary: Underlying platelet problems in majority of women with unexplained menorrhagia. Suggests need for screening for inherited blood disorders and platelet problems in women with menorrhagia.
Shankar 200448Systematic review

EL = 2−
11 studies included in reviewPopulation characteristics: Women; menorrhagia; screened for von Willebrand disease

Search undertaken on MEDLINE only using keyword search.

Country:
vWD as risk factor in menorrhagiaN/APrevalence of von Willebrand disease11 studies: 988 women with menorrhagia and vWD prevalence of 131 (13%, 95% CI 11 to 15.6%). Studies reported range from 5% to 24% of vWD.

4 studies from Europe, 5 from North America, 2 from elsewhere.

6 studies based on gynaecology outpatient clinics, 1 on coagulation clinic, 1 on administrative database, 2 on population study, 1 not stated.

Menorrhagia state based on history in 5 studies, PBAC in 2, Alkaline Haematin in 2, and not stated in 2.

vWF:Ag test only one used across studies, RiCof was second most common.

Cut-off for vWD varied between studies.

Different study designs and inclusion criteria probably account for differences between studies.
Funding source: Not stated

From: Evidence Tables

Cover of Heavy Menstrual Bleeding
Heavy Menstrual Bleeding.
NICE Clinical Guidelines, No. 44.
National Collaborating Centre for Women's and Children's Health (UK).
London: RCOG Press; 2007 Jan.
Copyright © 2007, National Collaborating Centre for Women's and Children's Health.

No part of this publication may be reproduced, stored or transmitted in any form or by any means, without the prior written permission of the publisher or, in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK [www.cla.co.uk]. Enquiries concerning reproduction outside the terms stated here should be sent to the publisher at the UK address printed on this page.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant laws and regulations and therefore for general use.

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.