|
Further refinements
in technology are needed to improve the sensitivity and specificity
of the Osmetech Microbial Analyzer.
Evaluation of a
novel diagnostic test for bacterial vaginosis: 'the electronic
nose'.
Hay P, Tummon A, Ogunfile M, Adebiyi
A, Adefowora A.
Int J STD AIDS 2003;14:114-118.
Summary:
Question
What are the sensitivity and specificity of the Osmetech Microbial
Analyzer for the diagnosis of bacterial vaginosis compared to composite
clinical criteria and Gram stain scoring, and what are the factors that
interfere with its performance?
Design
This article describes a blinded comparison of an electronic nose device,
the Amsel clinical composite criteria, and the Nugent Gram stain scoring
system for the diagnosis of bacterial vaginosis (BV) in women attending a
genitourinary clinic.
Participants
Consecutive women (n = 1037) presenting to the Genitourinary Medicine
Clinic at St. George's Hospital, London, as new cases, who required a
speculum examination and screening for genital infections, were tested.
Participants were excluded if they had taken antibiotics within the
previous month. The mean age was 29 years (range = 18 to 63). The ethnic
origin was 49% Caucasian, 34% Afro-Caribbean, 3% Indian sub-continent, 5%
Asian-other, and 9% other or unknown. The following genital infections
were diagnosed: 39% BV, 28% candida, 5.5% chlamydia, and 1.5% gonorrhea.
Description of Tests and Diagnostic
Standard
Endocervical swabs were collected for Gram stain and culture of N.
gonorrhoeae, and for detection of C. trachomatis (MicroTrack
EIA, Syva). Vaginal swabs were collected for Gram stain, wet mount, and
cultures for detection of T. vaginalis and candida, and for
detection of BV using Amsel clinical criteria, Nugent Gram stain scoring
system, and the Osmetech Microbial Analyzer™ (OMA).
The OMA electronic nose system uses a
vaginal fluid swab, placed in a sealed vial, and analyzed directly without
extraction. The amount of vaginal fluid specimen collected was determined
by weighing the swab before and after collection. The sample headspace was
passed over an array of conducting polymer sensors, each of which has
specific interactions with different volatile organic species. The
resistance of each sensor was measured and the percentage resistance
change calculated. Two principal component scores, PC1 and PC2, were used
to assess the system performance.
The first 373 subjects were used to
evaluate the optimum method of sampling. Among 664 subjects available for
assessing the OMA performance, 14 were excluded due to high PC2 scores.
All had either BV, candida, or both. Six additional subjects were excluded
due to unreadable Gram stain slides.
Main Outcome Measures
The sensitivity, specificity, positive predictive value (PPV), and
negative predictive value (NPV) of the OMA compared to Amsel clinical
criteria and to Nugent Gram stain scoring for the diagnosis of BV were
determined.
Main Results
The performance of the OMA, calculated using the optimum PC1 threshold of
-15, compared to Amsel clinical criteria and to Nugent Gram stain scoring
for the diagnosis of BV is shown in the table. Of 105 subjects who were
positive by OMA and negative by Nugent score, 19 had an intermediate
Nugent score of 5 or 6. The agreement between Amsel criteria and Nugent
scoring for the diagnosis of BV was 94% for 644 evaluable subjects. In a
multiple logistic regression analysis, delay of more than 12 hours before
measurement, candida on culture, intercourse within one day of testing,
and carry over effect with a sample following a BV positive one tended to
give more false positive OMA results. Swab weight did not affect the
performance.
Authors' Conclusions
The sensitivity and specificity of the OMA
were comparable to the results obtained with some commercially available
tests, but fall below the standards usually required for diagnostic tests.
Further refinements of the technology are needed to improve the
performance.
Source of funding:
None given
For correspondence: Phillip Hay,
Courtyard Clinic, St . George's Hospital, Blackshaw Road, London SW18 0QT,
UK. E-mail address: phay@sghms.ac.uk.
|
.gif){kind=small}