Electronic cancer detector to beat smear errors

An electronic cancer detector that can spot abnormalities in cervical smears automatically and could help put an end to the scandals that have beset the national cervical cancer screening programme is to be tried in Britain.

The computerised system, called Papnet, can process 70,000 smears a year and has been shown to reduce the error rate fourfold compared with human screeners. However, each detector costs pounds 500,000 and there are worries that it may prove unaffordable for the NHS.

Trials of the automated system are to begin in Liverpool and London following a series of scandals involving the misreading of smears by human screeners which have undermined the confidence of women in the cervical screening programme.

In the latest incident, Warwickshire Health Authority announced last week that it was re-examining the smears of 18,000 women after checks showed some that were abnormal had been missed. Earlier this month Frank Dobson, the Health Secretary, announced a national review of the breast and cervical screening programmes following disasters at the Exeter breast screening centre and at the Kent and Canterbury Hospital cervical screening laboratory, where five women developed cancer and 90,000 smears had to be re-checked.

Under the current system each cervical smear carrying between 300,000 and 500,000 cells is scanned by a human screener who can spot even one or two cells that are abnormal. However, the work, which takes eight minutes per slide, is tedious and repetitive and in 10 to 15 per cent of cases abnormal cells are missed.

The failure in detection has been blamed on the lack of experience of the smear checkers, who are poorly paid, while most labs are run by consultant pathologists with other interests who cannot give screening the attention it requires. However, errors in screening can creep in at any of a dozen points in the process, including in the GP's surgery.

The computerised Papnet system divides the slide into 128 sections and automatically selects the section containing the most suspicious cells. A human screener can then focus on that section to confirm the diagnosis. The Papnet's US manufacturers claim the error rate is only 3 per cent.

Dr Lesley Turnbull, director of the Royal Liverpool Hospital's screening laboratory, where the system will be tried out next month, said: "I hope this will be a way of increasing the sensitivity of the screening test. From what I have seen so far I think there will be a significant increase in the detection rate."

The trial, which is part-funded by the National Cervical Screening Programme, will put 20,000 smears through the Papnet system and through normal human screening and compare the results. A similar trial, funded under the NHS Research and Development programme, is being run at St Mary's Hospital, London.

Julietta Patnick, co-ordinator of the national screening programme, said 700,000 of the 4.5 million women screened each year were recalled for further checks although only 20,000 required treatment. "The Papnet system may well increase the sensitivity of the screening test and reduce the numbers recalled unnecessarily. That could be how we pay for it," she said.

The Papnet system has been available privately at a cost of pounds 45 a smear but the aggressive marketing strategy used to sell it drew criticism from doctors and the cervical screening programme.

Where it has all gone wrong

1993: Inverclyde Royal Hospital, Scotland. 20,000 smears re-checked after woman developed cervical cancer. Examination of earlier smears taken from the woman revealed abnormalities which should have been spotted.

1993: Gateshead, Tyne and Wear, and East Sussex. 800 women recalled for repeat smears after two GPs were found to be using an inappropriate smear test technique. One was using a gloved finger instead of the specially designed wooden spatula.

1994: Blackpool. 300 women called for repeat smears after being told their GP had used the wrong technique.

1994: Merthyr Tydfil. 18,000 smears re-checked at Prince Charles Hospital after unusually low proportion of abnormal smears was spotted.

1996: Gorleston, Norfolk. 9,000 smears re-checked at the James Paget Hospital after screeners who took slides home to examine made a series of errors. One woman screener was found to have misread 357 of the 10,000 smears she examined over three years.

1996: Kent and Canterbury Hospital. 90,000 smears re-checked after five women developed invasive cancer and over 300 found to be in need of urgent treatment. Described as the worst screening scandal, an inquiry found management failings and problems at every level from the screeners to the consultants.

1997: Warwickshire. 18,000 smears re-checked after low proportion of abnormal smears was detected. Monitoring disclosed 16 out of 500 smears had been misread.