Pointwise univariate linear regression of perimetric sensitivity against follow-up time in glaucoma

The authors compared pointwise univariate linear regression (ULR) of sensitivity against follow-up as an indicator of visual field progression with that of the corresponding ULR of mean deviation (MD) and with the Glaucoma Change Probability (GCP) analysis. The authors determined the influence of the number and sequence of prior examinations on the slope of the pointwise function.

Univariate linear regression was undertaken at each stimulus location on the arbitrarily assigned left eyes of 38 patients with glaucoma examined with the Humphrey Field Analyzer Programs 30-2 or 24-2 (stimulus size III, Humphrey Instruments Inc, San Leandro, CA). The mean age was 59.0 years (standard deviation [SD] = 12.9), the mean number of fields per patients was 12.0 (SD = 2.8), and the mean duration of follow-up was 6.0 years (SD = 1.6).

Four patients showed statistically significant MD slopes. Of the 34 patients exhibiting a nonsignificant MD slope, 15 exhibited clusters of at least two contiguous progressing locations. Less than half of these locations were designated as progressing by GCP. The GCP detected less than one third of the locations considered progressing by ULR for the last six fields in the series: this was attributed to the nonlinear nature of the decline in sensitivity.

The degree of agreement between the outcomes of ULR and GCP was dependent on the quality of the collected data, the magnitude of the baseline sensitivity, the extent and type of the subsequent visual field progression, and the position of the fields within the examination series. Good agreement was illustrated at those locations where the deterioration fell outside the limits of expected variability in stable glaucoma.