Detection and prediction of visual field progression has been recognized as a crucially needed development for glaucoma management and care. Characterization of visual field progression will allow for directed follow-up of high risk eyes, early detection and treatment of glaucoma, better potential for slowing progression of diseased eyes, and a protocol for delaying visual field damage. As a secondary outcome, it will also provide a justifiable and valid progression outcome measure for clinical trials and studies of glaucoma. However, it is equally well recognized that characterization of progression is one of the most challenging aspects of glaucoma research and clinical evaluation but perhaps a finding with high impact on management strategies/philosophies. The primary aims of this research project are to develop clinical indicators and prognostic factors for detecting and predicting glaucomatous visual field progression.
These aims will be met through applications of novel classification and regression tree (CART) methods to data from the perimetry and psychophysics in glaucoma (PPIG) study. The PPIG study consists of 168 individuals with moderate to high risk ocular hypertension or early glaucoma followed for up to eleven years with annual visual field examinations. The data includes a measure of progressive glaucomatous optic neuropathy (pGON), standard automated perimetry visual field test patterns and summaries, optic disc summaries, and other clinical and sociodemographic observations. The new CART methods developed account for correlations between fellow eyes and spatial variability in the temporal series of visual field measurements in assimilating visual field data. The CART analyses thus take full advantage of the PPIG study data in creating glaucomatous visual field progression classification systems, definitive measures for detecting visual field progression from follow-up visual fields and for predicting visual field progression from baseline observations.
We aim to develop progression classification systems for glaucoma, a common blinding eye disease, that have straightforward clinical interpretation, being a diagnostic indicator of progression and progression severity. The progression indicators will be based on clinical data commonly collected in practice and are thus easily incorporated into mainstream ophthalmic testing equipment and clinical risk calculators. The novel statistical methods to be developed could also serve as general data mining tools for public health researchers interested in detecting or predicting a given outcome (typically disease status) by using data collected over time, for example the results of tests commonly administered at periodic health physicals.
| Levine, Richard A; Fan, Juanjuan; Strickland, Pamela Ohman et al. (2012) Frailty Modeling via the Empirical Bayes Hastings Sampler. Comput Stat Data Anal 56:1303-1318 |
| Natarajan, Loki; Pu, Minya; Fan, Juanjuan et al. (2010) Measurement error of dietary self-report in intervention trials. Am J Epidemiol 172:819-27 |
