Monday, 12th May, 2014, 11.00-12.00

Regression models for categorical responses: an application to the interaction of different chronic comorbidities in HIV positive patients

Multimorbidity is defined as the co-occurrence of two or more chronic medical conditions in one person. Multimorbidity is turning into a major medical issue for both individuals and health care providers. It is well-known that multimorbidity correlates with age and, furthermore, that HIV-infected patients experience an increased prevalence of noninfectious comorbidities, compared with the general population. It has also been hypothesized that such increased prevalence is the result of premature aging of HIV-infected patients.

Guaraldi et al. (2011) investigated the effect of HIV-infection on the prevalence of five noninfectious chronic medical conditions from an Italian dataset obtained from a cross-sectional retrospective case-control study. More specifically, Guaraldi et al. (2011) analysed the effect of HIV-infection by means of logistic regressions on univariate outcomes. However, the analysis of single responses is not sufficient becasue multimorbidity is characterised by complex interactions of co-existing diseases and to gain relevant insight on the role of HIV-infection on multimorbitiy it is necessary to use a multivariate approach aimed to investigate the effect of HIV on the interaction of different chronic conditions.

In regression models for categorical responses a linear model is typically related to the response variables via a transformation of probabilities called a link function.  We present an approach that is based on the connection between two different links: the log-mean linear link and the Moebius link. In our framework, the interpretation of the effect of covariates on the interaction of responses is straightforward.

This is joint work with Monia Lupparelli, University of Bologna.

Speaker: Alberto Roverato, Professor of Statistics at the University of Bologna (Italy)

Room Xipre (seminar 173.06-183.01), PRBB.