Purpose: To determine whether or not there are systematic differences in the areas of optic discs as measured by different machines using different measurement algorithms and whether racial or gender differences exist in optic disc area measurements.
Methods: We examined the results of twenty-three published studies on the size of normal optic discs of various patient populations. Studies differed in the type of instrument and method used to measure optic disc area, and the number, age, race and gender of subjects examined. Noticing that different machines exhibited statistically significant systematic differences in optic disc sizes of comparable populations, we computed a "normalization" factor for each machine based on these mean differences. Applying this normalization factor to the results, we then re-examined the differences between racial and gender groups.
Results: By comparing the results of mean optic disc areas of different racial groups made with different machines, and normalizing results according to those of the Zeiss fundus camera, we found the normalization factors for the following machines to be, Zeiss fundus camera: 1 (by definition), Rodenstock Optic Disc Analyzer (RODA): 1.51, Topcon fundus camera: 1.04, Heidelberg Retina Tomograph (HRT): 1.15 and TopSS scanning laser ophthalmoscope: 1.29. That is, to bring the results of area measurements made with a RODA machine in line with those made with a Zeiss fundus camera, one should multiply the former by the factor 1.51. Using the normalized results, we confirmed the findings of previous authors that the optic disc areas of black subjects were statistically significantly larger than those of white subjects (n-weighted mean effect = 0.556 +/- 0.142 S.E., n = 5). Further, the meta-analysis of various racial populations from five studies shows that males have significantly larger discs than females (n-weighted mean effect = 0.151 +/- .055 S.E., n = 9).
Conclusion: Different machines and techniques give different results when populations of similar racial composition are measured. We recommend applying the above normalizing factors when comparing studies that employ different instruments.