Authors' recommendations: A systematic review in 2009 found three robustly designed studies showing no significant difference between image-assisted and manual reading of liquid based cytology (LBC) slides in the number of true cases of CIN2+ identified at any threshold. A systematic review found three robustly designed studies showing inconsistent results regarding false positives for CIN2+. Using high grade thresholds: two studies found that image-assisted reading generated significantly fewer false positives than manual reading of LBC slides, and one study found no significant difference. There were no significant differences in false positives for CIN2+ using low grade thresholds in the two studies which examined this. In 2011, the MAVARIC (Manual Assessment Versus Automated Reading in Cytology) trial found that image-assisted reading with ThinPrep® was associated with significantly reduced sensitivity for the detection of CIN2+ pathology compared with manual reading, but there was no significant difference between the two methods for detection of CIN3+. The Scottish Cervical Cytology ThinPrep® Imager Feasibility Study found similar levels of sensitivity and specificity between image-assisted and manual reading for the detection of all lesions. In phase II only, there was some indication of improved sensitivity in the detection of high grade lesions with image-assisted reading (statistical significance not reported). Differences in screening protocol, study design, reference standard, cytologist training and outcome measure make comparison between studies difficult. Productivity is improved with image-assisted screening. Cost-effectiveness modeling in the MAVARIC trial predicted that automated LBC would dominate manual LBC, based on estimates using aggregated data from two automated devices: ThinPrep® and BD SurePathTM. Automated reading was associated with cost savings, a small gain in quality adjusted life years (QALYs) and a combined incremental cost-effectiveness ratio (ICER) of -£7,592 per QALY gained, compared with manual reading, of LBC slides. However, trial authors urged caution in the interpretation of this result.

Project Status: Completed

Year Published: 2011

URL for published report: http://www.healthcareimprovementscotland.org/our_work/technologies_and_medicines/earlier_evidence_notes/evidence_note_34.aspx

English language abstract: An English language summary is available

Publication Type: Not Assigned

Country: Scotland, United Kingdom

Organisation Name: Scottish Health Technologies Group

Contact Address: Scottish Health Technologies Group, Delta House, 50 West Nile Street, Glasgow, G1 2NP Tel: 0141 225 6998

Contact Name: his.shtg@nhs.scot

Contact Email: his.shtg@nhs.scot

Copyright: Healthcare Improvement Scotland