Estimate the causal treatment effect for subjects that can adhere
to one or both of the treatments. Given longitudinal data with missing
observations, consistent causal effects are calculated. Unobserved potential
outcomes are estimated through direct integration as described in:
Qu et al., (2019)