Treatment-Confounder Feedback

Treatment-confounder feedback occurs when:

1. Treatment $A_{k-1}$ affects confounder $L_k$
2. Confounder $L_k$ affects treatment $A_k$
3. Confounder $L_k$ also affects outcome $Y$

Example: Antiretroviral therapy (ART)

• ART affects CD4 count ($L$)
• CD4 count determines next ART decision
• CD4 count affects survival

This creates a feedback loop that standard methods cannot handle.