Abstract
Purpose: :
To demonstrate how a previously published statistical eye model can be used to reconstruct the biometric data used by Sölve Stenström for his PhD thesis (1946), but that was lost over time.
Methods: :
Stenström’s thesis gives the average and standard deviation values, as well as the prevalence for the following parameters: refraction, axial length, anterior chamber depth, corneal curvature and crystalline lens power. He also included the correlations between all these parameters for a population of 1000 right eyes (20-35 year old, 685 males, 315 females). Assuming that all parameters are normally distributed, we have demonstrated that this information is sufficient to reconstruct the original data by means of a statistical eye model described in a previous paper.
Results: :
The normality of Stenström’s prevalence data could be verified for all parameters except refraction by means of QQ plots. Hence refraction was not included in the statistical model directly, but calculated from the data generated by the model. The averages and standard deviations of the reconstructed data were found to correspond very well with the original values reported by Stenström, with an exception of the standard deviations of the refraction and the axial length. These were found to be significantly smaller in the generated data, probably due to cases of extreme ametropia which were present in Stenström’s data, but not in the model. If eyes with an ametropia larger than -8D or +8D are excluded, the differences between the standard deviations reported by Stenström and those generated by the statistical eye model reduced to non-significant levels.
Conclusions: :
Using statistical eye modeling the original data set could be reconstructed reliably within a refractive range of -8D to +8D. Given the high quality and completeness of Stenström’s measurements, this reconstructed data could help in the study of how ocular biometry has changed throughout the past 70 years.
Keywords: computational modeling • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • refraction