Graefe's Arch Clin Exp Ophthalmot (1989) 227:513-517 Graefe's Archive for Clinical and Experimental Ophthalmology © Springer-Verlag1989 The distribution of normal values in automated perimetry* C. Rutishauser, Josef Flammer, and A. Haas Universit/its-Augenklinik, Mittlere Strasse 91, Ch-4056 Base1, Switzerland Abstract. From 354 visual fields of 137 normal subjects, various components of variance were calculated separately for each test location of the Octopus automated-perimetry program J0. The method of component analysis of variance was used. The following components were analyzed: interin- dividual variation (variation of visual-field measurements in different subjects), long-term fluctuation (variation of dif- ferent visual-field measurements in the same subject), differ- ences between the right and left eyes and fluctuation within one visual field test in one subject, i.e., short-term fluctua- tion. The results show increased variations at the center relative to the paracentral area and a slight increase with eccentricity. Introduction Automated static perimetry provides an efficient and accu- rate quantitative measurement of differential light sensitivi- ty. This renders possible the detection of shallow defects, which in turn enables the detection of early damage. This is of clinical relevance for several diseases such as glaucoma and pituitary tumors, among others. A prerequisite for the detection and definition of shallow defects is knowledge of normal values [4] and their distribution. Normal values have been an essential component of Oc- topus automated perimetry since its inception. These nor- mal values have been confirmed and specified in a further study [10]. The reproducibility of the results of quantitative perimetry has previously been studied with the help of man- ual perimetry [2]. The introduction of automated perimetry, however, enabled a more detailed analysis of the fluctua- tions in perimetric outcomes. Bebi6 et al. [1] were the first to describe the short- and long-term components of fluctuation. Furthermore, they separated spatially correlated from spatially non-correlated long-term fluctuations, i.e., those long-term fluctuations that depend on the area examined in the visual field and those that are independent of test location. In another paper [31, these were referred to as the homogeneous and heteroge- neous components of the long-term fluctuation. The size of the fluctuation and the factors influencing it have been * This study was supported by the Swiss Science Foundation (3'790-0.84) Offprint requests to: J. Flammer studied in a number of investigations [5, 8]. It has been reported that glaucoma suspects, for example, have a larger fluctuation in differential light sensitivity than do normal subjects [6, 7, 9, 16 18]. The purpose of the present study was to analyze sepa- rately for each individual test location the magnitude of the following quantities in normal subjects: the short- and long-term fluctuations, the interindividual variation, and the differences between the right and left eyes. Subjects and methods This analysis is based on a pool of visual-field data from normal subjects who were examined using Octopus au- tomated-perimetry program J0 [13]. This program mea- sures twice the differential light sensitivity at 49 test loca- tions, up to an eccentricity of 26 ° (horizontally, +_ 21°; verti- cally, __ 15°). The data consisted of 354 visual-field measurements in 240 eyes of 187 normals between 12 and 81 years of age. After the overall distribution had been calculated using the entire pool of data, various examinations were selected for the analysis of the various components contributing to the total distribution (Table 1). For the calculation of the interindividual variation and the short-term fluctuation, we selected one eye per individ- ual. The total number of individuals included was 137. For the calculation of the long-term fluctuation, we selected one eye (right or left) of all individuals who underwent two visu- al-field measurements of the same eye on different days (44 subjects). For the calculation of the differences between right and left eyes, we selected 47 individuals who underwent visual- field measurements of both eyes. To obtain a sufficiently large number of subjects, we used the first measurement for this analysis. A mixed model for the analysis of variance was used to sort out the different components [11]. The analysis was performed separately for each test location. The following model of variance was applied: yjkl= # + Aj+ Pk+ LI+ A " P~k+ P. Lkl + A ' P ' L j k l + A GE + E~k~, where Yjkl represents the individual threshold at each test location for eye j of patient k at session l; # represents the overall mean value for the corresponding test location;