Cochrane reviewer Mark Jones and intensive care specialist Robert Fowler using simulation and an example demonstrate the impact of immortal time bias on the observational datasets used in Tamiflu influenza survival studies. The use of inappropriate statistical methods shows an effect on mortality which disappears when time-dependent exposures are included as time-dependent variables in the analysis.

http://www.sciencedirect.com/science/article/pii/S0883944116302544