Video-on-Demand Library

This content is restricted to members of PSI. If you are already a member please login. To join PSI or to see all the benefits of membership click here.

03 June 2019

In influenza, a viral infection starts an immune response responsible for the observed symptoms. Although inter-individual differences in intensity and duration of symptoms can be partly explained by prior exposures and genetic differences in immunity (e.g. variation at the MHC gene complex), it is possible and desirable that the immune response also adapts to the level of viral infection over time (e.g. to prevent overreactions). Despite the ensuing adaptation between viral level and immune response, the literature shows difficulties to quantify this relationship. For example, Carrat et al. (2008) showed viral titer and symptoms score having similar trends but did not statistically quantify such association. It is notoriously difficult to appreciate such association by simple plots, e.g. time to resolution of symptoms against total viral load; more sophisticated statistical approaches are required. Here we compare two statistical approaches: one based on time-dependent Cox models (Collett 2015) and the other based on joint models (Rizopoulos 2012). The level of association between symptoms and viral load is quantified, and can be used to prognosticate disease evolution, in both approaches. However, joint models offer several advantages over Cox models, the most important of which is that they predict a smooth evolution of viral kinetics using all available longitudinal virology information, whereas in a Cox model, the level of viral infection is assumed constant between consecutive measures. Hence, joint models are proposed as the preferred statistical approach to quantify the association between viral kinetics and resolution of symptoms in influenza.

Related Documents

Part of Collection(s)

Topic(s)

Upcoming Events

Latest Jobs