Marginal structural model (MSM) is used to assess the effect of treatment in the presence of a time-dependent confounding variable. In this scenario treatment is provided to individuals at multiple time points and can be influenced by a variety of factors.Typical example is HIV whereby the type of treatment is dictated by CD4 count which itself is affected by past exposure. The MSM can handle this situation but assumes the effect of treatment (causal effect) is consistent across levels of the confounding factor (i.e., change in CD4 count). Can you think of a situation where the effect of treatment may be different as function of the time-dependent confounding factor?