The Royal Statistical Society GAS & RSS Merseyside local group
13:00-13:45 Dr Catey Bunce (Principal Statistician, NIHR BRC for Ophthalmology at Moorfields and the Institute of Ophthalmology, UCL)
Method comparison studies in ophthalmology
Ophthalmic researchers today are faced with a variety of methods of measuring ocular characteristics such as intraocular pressure, retinal nerve fibre layer thickness and macular thickness. It is not uncommon to find papers which report on whether or not such methods are interchangeable. It is also not uncommon to find errors in the statistical analyses in such papers or in the interpretation of results in such papers. This talk highlights some common misunderstandings in the analysis of such studies, with some applied examples, promotes the use of limits of agreement and questions why old habits are hard to change. An introduction to the Ophthalmic Statistics Group will be provided – this being a group of medical statisticians working across the UK and beyond who have come together because of a common interest in ophthalmic research and desire to raise statistical standards in ophthalmic research.
13:45-14:30 Dr. Gabriela Czanner (Lecturer in Biostatistics, Department of Biostatistics and Department of Eye and Vision, University of Liverpool)
Problems of dichotomisation in clinical studies
In ophthalmic clinical studies we often measure continuous variables (such as intra-ocular pressure, visual acuity, contrast sensitivity, etc). Much of medicine revolves around an implicit classification of individuals into diseased and non-diseased or into categories created by applying a cutpoint in the value of a continuous variable. for example a diagnosis of glaucoma might be confirmed by an elevated intra-ocular pressure (IOP) measurement (eg IOP > 21 mmHg). In clinical research, continuous variables may likewise be converted to categorical variables, assigning individuals to one of two groups. We discuss situations when this may be appropriate and situations when such dichotomisation has sever drawbacks such as loss of information, loss of statistical power and bias in estimated associations.
15:00-15:45 Miss Rachel Nash (Medical Statistician, Clinical Trials and Evaluation Unit, School of Clinical Sciences, University of Bristol)
Analysing data from randomised controlled trials with baseline and follow-up measurements – what is wrong with analysing change from baseline?
In clinical trials, continuous outcomes, such as intraocular pressure and visual acuity, are often measured both before treatment (ie at baseline) and after treatment. Having the baseline measurement allows us to account for the initial differences between patients, when comparing the outcomes of alternative treatments. Whether or not baseline measurements are accounted for in an analysis may impact on the results of a trial. Three common approaches to the analysis of clinical trial data when we have both baseline and post-treatment values, namely: analysis of covariance (ANCOVA), analysing post-treatment values only (ie ignoring baseline measurements), and analysing change scores (ie the difference between the post-treatment measurement and baseline measurement for each participant) will be compared, with recommendations.