Is Astigmatism Correlated with Myopia or Hyperopia?

Abstract

Abstract: : Purpose: Several papers (including Fulton et al., 1982 and Kay and Paterson, 1997) have suggested that myopia and astigmatism are strongly correlated. We examine this issue in three distinctly different populations. Methods: The three populations were the following: 1) The first 2160 patients of all ages that presented at our clinic in Guanajuato, Mexico starting in February 2000; 2) the 2778 patients (from England) reported in the Sorsby publications dated from 1957 through 1970; and 3) 456 unselected children from central Indiana refracted at the beginning of the first (G1) and fourth grades (G4). The data on the children were collected starting in 1995. All refractive errors of both eyes were converted to M, J0, and J45 power vector coefficients (Thibos et al., 1997) and submitted for Matlab software analysis. Significance was tested utilizing Hotelling’s T2 statistic. Results: The data from the right eyes will be reported below. In all cases data from the left eyes were essentially the same for all mean values reported below. The population means were statistically different from each other (p<0.001). This difference was due primarily to the mean sphere component. The mean M (sphere equivalent) was –0.22 D for the data from Mexico, +0.45 D for the G1 children, -0.08 D for the G4 children, and +1.16 D for the Sorsby data. To summarize the astigmatism of each population, two-dimensional 95% confidence ellipses were constructed with the J0 and J45 coefficients. The ellipses were all elongated in the J0 direction by approximately a factor two. The centers of the ellipses were close to the origin for the children not significantly different from each other (J0 = 0.04 D, J45 = 0.00 D for G1). The centers of the ellipses for the other two populations were shifted slightly in the direction of with-the-rule astigmatism in the J0 coefficients (0.12 D for Mexico and 0.14 D for Sorsby). The J45 coefficients were close to the origin (0.06 D for Mexico and 0.03 D for Sorsby. The J0 95% confidence limits for the children were 0.5 D at G1 and 0.6 D at G4. The Sorsby 95% confidence ellipse was slightly larger with the J0 limit at 0.85 D. Substantially more astigmatism was found in the Mexican population with the JO limit extending to 1.4 D. The correlation between M and astigmatism was found to be very low in the G1, G4 and Sorsby data accounting for only 0.01% of the variance in each case. The correlation of the Mexican subjects accounted for only 7% of the variance. Conclusion: The hypothesis that myopia is strongly correlated with astigmatism is not well supported by these data.

Keywords: astigmatism • refractive error development • clinical (human) or epidemiologic studies: bio