Page 46 - Screening for Cervical Cancer: Systematic Evidence Review

Chapter III. Results

Key Question 2

New Methods for Preparing or Evaluating Cervical Cytology

Key Question 2 (To what extent do new methods for preparing or evaluating cervical

cytology improve diagnostic yield compared to conventional methods?) is addressed

predominantly by direct comparisons of diagnostic tests. The majority of the literature identified

for Key Question 2 is based on archived laboratory specimens. These studies compare the

techniques being evaluated with the results of review by a panel of cytology experts. Most often,

these comparisons are conducted by subjecting specimens with a pre-selected mix of normal and

abnormal cytologic results, to review by the techniques under study. In general, discrepancies

between cytology reports were adjudicated by the expert panel masked to findings; rarely,

subsets of normal or concordant diagnoses were also reviewed by the panel.

Very few studies of new technologies are validated by concurrent or subsequent

colposcopy or histology of abnormal screening test results; even fewer include validation of

normal screening test results. This means that in almost all studies identified the sensitivity,

specificity, and predictive values of the technology cannot be directly assessed or compared with

the test characteristics of conventional cytology in the same population.

Our search identified 196 articles on the types of technology we wished to review. We

excluded 143 of these articles at the time of abstract review because they did not meet basic

inclusion criteria (e.g., were commentaries, were based on experimental laboratory systems, did

not have human subjects). Forty-eight full articles were retrieved; these included 23 that were

found not to meet basic inclusion criteria (commentary and reviews). Of the remaining 25

articles, four were relevant new articles not previously abstracted and summarized in the AHCPR

46