Computerized Cytology Tops Conventional in Detecting Cervical Abnormalities

SYDNEY, Australia, June 29 -- A computerized system for reading liquid-based cytology slides outperformed conventional cytology for detecting cervical abnormalities.

The automated slide reader classified more specimens as abnormal (7.4% versus 6%) and detected more cases of grade 1+ cervical intraepithelial neoplasia (2.8% versus 2.0%), according to a study reported online in the British Medical Journal.

Computerized imaging of specimens also resulted in significantly fewer unsatisfactory slides compared to conventional cytology (1.8% versus 3.1%, P

The primary outcome measure was the accuracy of slides for detecting squamous-cell cervical lesions. Secondary outcomes were the proportion of unsatisfactory slides, distribution of squamous cytological classifications, and accuracy for detecting glandular lesions.

Among the findings, the computerized imager was significantly more likely than conventional cytology to classify slides as:

• Atypia (Odds Ratio 1.8, P=0.018)
• Atypia with signs of human papillomavirus (OR 1.65, P

For example, in the May 21 issue of the British Medical Journal, Italian investigators reported finding no difference in accuracy between computerized and conventional cytology for detecting CIN2+.

Moreover, when Dr. Davey and colleagues reviewed 46 studies comparing the two cytology techniques, they found no difference in the proportion of unsatisfactory slides, which they reported last year in The Lancet.