Abstract
Kerato (cornea) and Konos (cone) are two Greek words from which the term keratoconus (KC) is derived. The literal translation, that is, cone-shaped cornea, is the fundamental characteristic that defines the disease. Most research papers written on keratoconus begin by describing the disease as a bilateral, asymmetrical, non-inflammatory thinning of the cornea which leads to irregular astigmatism, corneal bulging and eventually scarring. Keratoconus is a multifactorial disease and much research has been done on many different aspects in an effort to better understand its complexities. Keratoconus affects the anterior segment of the eye especially the corneal thickness and curvature which directly impacts the corneal power and refractive state of the eye. In the majority of previous research on these particular aspects, data is inadvertently analysed partially or incompletely. For example, the axis or meridional orientation of keratometric measurements (or dioptric power) are sometimes omitted and/or only the spherical equivalent of dioptric power is used in the statistical analysis. The aim of this study is to investigate the intra-session and inter-session short-term variation of keratometric and refractive behaviour in patients with keratoconus holistically using multivariate statistical methods. Such methods rely on linear algebra and matrices for analysis of symmetric dioptric power. An analysis of variation of corneal pachymetry is also included in an effort to understand further its relation to the two previous parameters and also implications regarding the progression of the condition (KC). Forty successive measurements were taken with both the Oculus Pentacam and Nidek auto-refractor on 28 eyes with KC and 28 healthy control eyes to investigate intra-participant and inter-participant short-term variation for one measuring session. A second set of 40 consecutive measurements were obtained also for those participants who could attend a second measuring session one week later for eventual inter-session analysis. The analysis of data for the three parameters of interest (central corneal power, auto-refraction and pachymetry) was executed with the use of a combination of multivariate and univariate methods. This dissertation thus encompasses a comprehensive evaluation of short-term variation (both intra-session, intersession, intra-participant and inter-participant) of keratometric and refractive behaviour and pachymetry in KC and healthy eyes using a combination of multivariate and univariate statistical methods and to the best of the author’s knowledge, this is the first study of its kind. Multivariate methods, which include statistical tools such as three-dimensional stereo-pair scatter plots, 95% ii distribution ellipsoids, 95% confidence ellipsoids, comets, polar profiles, variance-covariance matrices and hypothesis testing are explained and applied to trivariate keratometric and refractive data over one or two sessions. Univariate data (such as variances and ellipsoid volumes) are analysed by conventional univariate methods, which include descriptive statistics of central tendency such as means and medians, dispersion (mainly standard and quantile deviations and variances), box-and-whisker plots and line plots. Traditional univariate methods, including Bland-Altman plots, are used to analyse pachymetry.
D.Phil. (Optometry)