Gardiner and Crabb reported in the British Journal of Ophthalmology1 that, when applying point by point regression models to evaluate perimetric progression, little information is gained by performing more than three perimetric tests a year because of threshold fluctuation. Increasing this frequency improves sensitivity but worsens specificity. Their criterion allows detection of progression in the range of 1–2 dB/year.

Threshold fluctuation may be reduced by applying mathematical filtering procedures, averaging the value of each threshold with those surrounding it or with those functionally related to it.2^–4 This procedure conserves most of the topographic information and may …