PART TWO
Diagnostic Cytology
developing CIN grade 3 were observed between women with
only one initial screening and women rescreened within the
3-year interval. The sensitivity of screening was calculated with
and without colposcopy. When screening sensitivity was assu-
med to be a constant, the sensitivity of the cytology-colposcopy
combination was 0.71, with 0.64 for cytology alone. When sen-
sitivity was assumed to decrease in subsequent screening rounds,
the complex sensitivity was calculated to be 0.60, with 0.50 in
subsequent rounds and 0.52 for cytology alone. We calculated
sensitivity of screening for the detection of severe dysplasia,
carcinoma in situ, and invasive cancer both separately and as a
group.126 For those women who had a tissue diagnosis of severe
dysplasia, carcinoma in situ, and invasive cancer and who had
participated in the screening program, the cytologic diagnosis of
the smear, taken up to 48 months earlier, was recorded. When
severe dysplasia, carcinoma in situ, and invasive cancer was con-
sidered together in the evaluation of false-negative diagnoses
after 24 and 36 months, sensitivity figures were 94 and 89%,
respectively. When only carcinoma in situ and invasive cancers
were considered, sensitivity figures after 24, 36, and 48 months
were 95, 92, and 89%, respectively. Sensitivity was relatively
high, being 99, 98, and 97%, respectively, when only invasive
cancers were considered to be falsely negative in the analysis of
diagnoses after 24, 36, and 48 months. If an epithelial abnor-
mality is diagnosed in a woman who previously had a negative
test result, two possibilities must be considered: (1) the epithe-
lial abnormality did not exist when the former negative smear
was taken, or (2 ) the epithelial abnormality did exist but was not
recognized (screening or interpretation error) or was not present
in the cell sample (sampling error). To calculate the sensitiv-
ity of cervical cytology, information was to be obtained on the
number of false-positive results and false-negative diagnoses.
Almost all of these studies evaluated sensitivity and specificity
(e.g. the validity of the test).72,79,127,128,258 Much less research
has been done on the reproducibility of light microscopic
diagnosis.270,271
Observer Variability
Observer variability is the variation in scoring by observers. It
can be divided into intraobserver variability, interobserver vari-
ability, and the variability due to inaccuracy of the assessment
system, such as changes within the microscope or fading of
staining.
Intraobserver variability is defined as the difference in scor-
ing by the same observer when evaluating the same specimen
on two separate occasions, correcting for the inaccuracy of the
system. Interobserver variability is defined as the difference
in scoring between observers.271,272 The observer variability in
cervical cytology has only rarely been evaluated.173,270-274 Intra-
observer variability proved to be an important factor in incor-
rect diagnoses. Intraobserver variability was found to be rather
inconsistent. Overdiagnoses as well as underdiagnoses were
made by all observers without evidence of a consistent pattern
when reviewing previously screened cases. Inconsistency was
not significantly reduced with longer experience in cytodiagnos-
tics.173,270-272 It was found that 9.1% of smears with false-negative
and 52.9% of smears with false-positive diagnoses were correctly
diagnosed at a second screening by the same observer. Intra-
observer variability was less than 17% when the same observer
screened the same smear twice. However, average intraobserver
variability differed considerably for individual observers. The
intraobserver variability seemed only slightly influenced by the
years of experience in cytopathology. Interobserver variability
also showed considerable differences between observers. The
observer's years of full-time experience in cytopathology had
a strong influence on the grading of squamous and squamous
metaplastic abnormalities. Evans and co-workers, in one of the
few studies on observer variability, suggested that cytodiagnosis
in itself was not an appropriate field for quality control because
it is essentially an expression of opinion.273 From our studies, we
have deduced that intraobserver variability is the main cause of
false diagnoses.270-272 When wrongly diagnosed, severe epithelial
abnormalities are more often underestimated than completely
overlooked. Apart from training in cytopathology, the estab-
lishment of laboratory protocols for multiple screening of even
minor abnormalities and the institution of a well-supervised
protocol for repeat examinations after cytologic diagnoses of
epithelial abnormalities of mild and moderate severity seem to
be the most effective means of reducing the number of severe
epithelial abnormalities that remain undetected at cervical cyto-
logic screening.122,124,258
Quality Control
The cytologic report should encompass the following items21,70,
74,75122,124,128,275,276:
• A statement about the adequacy of the specimen,
including an explanation of the problems encountered
for less than fully satisfactory specimens and a deter-
mination of whether a repeat specimen is necessary;
• A descriptive diagnosis comprising the presence and
character of any inflammatory changes, the expected
histopathologic change in the squamous or squamous
metaplastic cervical mucosa, and changes in columnar
cells from the endocervical mucosa or an abnormality
related to the endometrium; and
• A recommendation for further action to be taken on
the basis of the cytologic diagnosis.
Quality control by the laboratory seems the best way to reduce
sampling and screening errors.122,124,128,275 Gay and associates
found that for all types of malignancy, the majority of errors were
due to poor sampling.277 They found an overall false-negative
rate of 20% for invasive processes. They further stated that if all
the errors made by laboratory personnel could be eliminated,
in their material the false-negative rate in the presence of malig-
nancy could have been reduced from 20 to 12%, the latter value
simply due to inadequate cell samples. Reporting inadequate
smears is an important step in ensuring the quality of laboratory
performance. An adequate cytologic sample is a smear that dis-
closes abnormalities of epithelial cells from the uterine sample
and thus triggers further action.257 Such a sample includes a suf-
ficient number of cells representative of the area sampled, fixed
and stained in a manner that allows interpretation.278
In an adequate cervical smear, endocervical columnar cells,
squamous metaplastic cells, and squamous cells should be
present. The cytopathologist should evaluate whether the speci-
men submitted is of sufficient quality to enable an adequate
diagnosis. This evaluation encompasses the cellular composition
of the sample, the quality of the cellular material, and technical
aspects of the smear, such as staining, cellularity, admixture of
blood and inflammatory cells, and adequacy of the specimen in
relation to the clinical information. If the specimen is less than
204
previous page 205 ComprehensiveCytopathology 1104p 2008 read online next page 207 ComprehensiveCytopathology 1104p 2008 read online Home Toggle text on/off