Multiple count sampling of goblet cells in microscope high-power fields using conjunctival impression cytology.

BACKGROUND: Published studies indicate that assessments of goblet cell density using conjunctival impression cytology has provided very variable results, but the reasons for this are unclear. Systematic analyses of the sources of variability are required. METHODS: From 20 healthy young adults, conjunctival impression cytology specimens were obtained using a supported filter unit applied to the superior bulbar conjunctiva. The filters were stained with Giemsa and 10 non-overlapping, randomly selected high-power field images were obtained from each specimen and the numbers of goblet cells per high-power field counted. RESULTS: From all 200 high-power fields assessed, the numbers of goblet cells ranged from zero to 74, with an overall mean value of 11.6 ± 14.8 per high-power field. From each successive set of 10 microscope field images from all individuals, the average number of goblet cells ranged from 23.2 in the first high-power field that obviously included numerous goblet cells down to 6.2 per high-power field. As the outcome from multiple counts/individual was systematically increased, these averages progressively decreased from 23.2 to 11.6 per high-power field, and while the standard deviation values also progressively declined (from 7.9 to 5.5 per high-power field), the relative variability (as the co-efficient of variation) did not, and increased to averaged values of over 100 per cent. CONCLUSIONS: These analyses indicate that there is a benefit of making multiple counts of goblet cells from different high-power fields, but that there is no obvious benefit of using more than five to seven high-power fields for any particular specimen.