A0242
Title: Visualizing departures from symmetry: A study on cardiovascular risk among patients with diabetes
Authors: Rosaria Lombardo - University of Campania (Italy) [presenting]
Eric Beh - University of Wollongong (Australia)
Antonio Ceriello - IRCCS MultiMedica (Italy)
Giuseppe Lucisano - Center for Outcomes Research and Clinical Epidemiology (Italy)
Francesco Prattichizzo - IRCCS MultiMedica (Italy)
Antonio Nicolucci - Center for Outcomes Research and Clinical Epidemiology (Italy)
Abstract: Sometimes, the same categorical variable is studied over different time periods or across different cohorts at the same time. In the case of studying the symmetry of a variable, Bowker's chi-squared statistic, presented in 1948, provides a simple numerical means of assessing symmetry for squared contingency tables. We analyse how row and column categories observed on different occasions deviate from the null hypothesis of perfect symmetry. In doing so, a generalization of Bowker's statistic for three-way squared contingency tables is provided. We focus our attention on studying the asymmetry that exists between glycated haemoglobin (HbA1C) variability (in subgroups of subjects with Type 2 diabetes who have a mean HbA1c <53 mmol/mol and > 53 mmol/mol) and cardiovascular complications (myocardial infarction, stroke, coronary revascularization/reperfusion procedures, peripheral revascularization procedures) when additional information is available on the subjects (such as their gender, weight and cholesterol level). This examination of the asymmetry shows the importance of investigating how similar/different, the cardiovascular complications are in cohorts of subjects with varying HbA1C levels that are observed at different occasions. We present a method of analysing and visualizing any departures from symmetry using a variant of correspondence analysis where Bowkers chi-squared statistic is used as its numerical foundation.
