OBJECTIVE: To accurately determine whether there are any features of an abnormal cervical smear that predispose to the production of a false negative report, in order to gain insight into why false negative reports are issued, and to establish whether there are steps that can be taken to reduce them. DESIGN: A quantitative retrospective analysis using the AxioHOME microscope of the number, size, and spatial distribution of abnormal cells in a set of 50 slides comprising a mixture of false negative and true positive cervical smears. SETTING: Five different cytology laboratories in the United Kingdom. RESULTS: False negative smears were found to be quantitatively different from true positive smears. They contained significantly fewer abnormal cells (median number of abnormal cells for false negatives = 173, median number of abnormal cells for true positives = 1712; p < 0.004), and these were more likely to be unevenly distributed on the slide. It was possible to predict with a high degree of accuracy whether a smear was a false negative by analysing number and distribution alone (kappa = 0.57). CONCLUSIONS: False negatives are quantitatively different from true positive cervical smears. This has important implications for quality assurance in the UK cervical screening programme. More consideration needs to be given to the effectiveness of existing quality assurance measures, which need to be tailored to the preferential detection of this type of abnormal cervical smear.
OBJECTIVE: To accurately determine whether there are any features of an abnormal cervical smear that predispose to the production of a false negative report, in order to gain insight into why false negative reports are issued, and to establish whether there are steps that can be taken to reduce them. DESIGN: A quantitative retrospective analysis using the AxioHOME microscope of the number, size, and spatial distribution of abnormal cells in a set of 50 slides comprising a mixture of false negative and true positive cervical smears. SETTING: Five different cytology laboratories in the United Kingdom. RESULTS: False negative smears were found to be quantitatively different from true positive smears. They contained significantly fewer abnormal cells (median number of abnormal cells for false negatives = 173, median number of abnormal cells for true positives = 1712; p < 0.004), and these were more likely to be unevenly distributed on the slide. It was possible to predict with a high degree of accuracy whether a smear was a false negative by analysing number and distribution alone (kappa = 0.57). CONCLUSIONS: False negatives are quantitatively different from true positive cervical smears. This has important implications for quality assurance in the UK cervical screening programme. More consideration needs to be given to the effectiveness of existing quality assurance measures, which need to be tailored to the preferential detection of this type of abnormal cervical smear.