To the Editor.
—The article in the July Archives by Aisen et al1 is misleading in its statistical analysis. It was stated that the hematocrit reading and retinal venous length correlate for all subjects significantly, as shown on a line of best fit drawn through a venous v hematocrit scattergram. The line of best fit as shown on the figure could lead the reader to wrongly conclude that the hematocrit reading is correlated with venous length for both anemic patients as a group and for control subjects as a group.We have reanalyzed the data presented in Table 1 and the Figure, and we find that within each group, there is no venous length-hematocrit correlation (r = −.253 for patients and −.127 for controls). Since the statistically significant difference in venous length between anemic patients and the control group is not accompanied by a correlation between the hematocrit reading and
Fletcher ME, Farber MD, Cohen SB, Jampol LM. Retinal Abnormalities Associated With Anemia. Arch Ophthalmol. 1984;102(3):358. doi:10.1001/archopht.1984.01040030278017