An evaluation of the quality of improvement potential of computer assisted screening technology within a cervical cancer screening programme

Abstract

Between 2006 and 2011, three studies - known as CAESAR (Computer Assisted Evaluation, Screening And Reporting) were carried out to evaluate Computer Assisted Screening (CAS) with manual primary screening to current procedures and protocols operated by the Welsh cervical screening programme – Cervical Screening Wales (CSW). A total of 45,317 SurePath™ liquid based cytology (LBC) cervical screening samples were submitted for CAS within four Welsh Cytology laboratories as part of a multi-centre randomised controlled trial.

The CAS technology chosen was the Becton-Dickinson FocalPoint™ GS Slide Imaging System (FocalPoint™) technology and a comparative assessment was carried out between the slides scanned and categorised using this technology (n=45,317) and those primary screened (n=137,806) over the same period and reported using established Cervical Screening Wales protocols, with a histological outcome where appropriate.

This thesis investigated several potential areas where this technology can be applied, in an effort to identify the overall benefits of the technology to the cervical screening programme.

These areas include: • Rapid quality assurance screening • Comparison of manual to automated primary screening • No further review (NFR) reporting category • Evaluation of the automated detection of endocervical cells • Screener acceptance of the technology • The relationship of the FocalPoint™ quintile ranking facility to sample Human Papillomavirus (HrHPV) status.

In addition, an economic analysis was carried out to identify any benefits and potential savings that might be realised by Cervical Screening Wales. The results of the study areas that have been analysed indicate that:

• The timings for carrying out a rapid QA screen via FocalPoint™ are comparable to a manual rapid pre-screen/re-screen and the sensitivity of the technology was determined to be at least equivalent to the rapid QA screens currently employed in cervical cytology.

• FocalPoint™ as a primary screening tool is not as sensitive for high grade or all grades dyskaryosis as manual primary screening as evidenced by not meeting the current minimum NHS Cervical Screening Programme (CSP) standards. Furthermore, the interval outcome rates for FocalPoint™ primary screened samples were greater than those for manually screened samples, both at 2 and at 3 year intervals.

• The 3-year interval outcome rates for CIN 2+ and cervical pre-cancer are significantly lower for the FocalPoint™ NFR category than those for manual screening. Interval cancer rates were similar, indicating that the FocalPoint NFR™ was demonstrably superior to manual primary screening in terms of fewer false negative results.

• System calibration and operational monitoring of the FocalPoint™ technology is vital for correct operation and optimised performance of the technology. This study highlighted on a hitherto unprecedented behaviour of the NFR reporting technology and brought about a significant revision of the manufacturer’s operating, calibration and monitoring procedures. The revised protocol was communicated to the NHS CSP task and finish group producing the NFR guidance document and the updated LPCA calibration procedure incorporated into the guidance (Denton et al., 2013). Therefore, as a direct result of this study, a major change in practice benefited a very large population of women (mainly outside the UK and Ireland) who received an improved screening outcome as a consequence.

• The inter-laboratory detection rates of endocervical cells by FocalPoint™ are more consistent from laboratory to laboratory than those for manual screening and yet those detection rates are equivalent to those of manual screening. The technology is therefore equivalent to manual screening for the detection of endocervical cells and may be used for the quality assurance of sample takers in the same way as manual primary cytology screening.

• Screener perceptions and acceptance of the FocalPoint™ technology were positive with participating individuals mainly appreciative of the diversion offered by FocalPoint™ from manual primary screening. Respondent numbers were small, however, and further, more structured investigation and analysis should be undertaken to qualify these findings.

The outcome of the economic analysis initially indicated that FocalPoint™ was expensive to implement and therefore offered little advantage to manual screening in a working laboratory. Further analysis, considering insufficient staffing levels and the resultant requirement for backlog management indicated that FocalPoint™ NFR and rapid QA screening offered a viable alternative to overtime working at enhanced pay rates for staff and would be of use in situations where screening staff were difficult to recruit.

• Comparison of FocalPoint™ quintile ranking rates with HrHPV results on LBC samples provided unexpected results, however, the samples compared were from a specific cohort of women that had received recent treatment for high grade CIN and this might account for the results of the comparison. Numbers of samples compared were low (n = 124) and this is a factor for consideration, indicating that further work is required to evaluate the FocalPoint™ technology in conjunction with HPV primary screening.

• In conclusion, the BD FocalPoint™ GS imaging system offers several advantages that are worthy of consideration and implementation by laboratories offering a cervical screening service.