Suraj Kapoor - there are major disadvantages in a per-protocol analysis vs intention to treat. In very simple terms the two groups compared in a pe protocol analysis are not formed by randomisation alone - they are formed by random allocation to groups + whatever process led to them not following the protocol. The potential for bias is large. In most cases, the factors that may generate bias are unmeasured and so cannot be quantified. Conceptually intention to treat mimics the process of real-world treatment and so can estimate average treatment effects when a treatment is used. Of course in reality the world is more complex than that and per-protocol analysis has a role to estimate effects in treated individuals. However I would say that the itrinsic bias means any caual inference must be weaker and if possible it should be secondary to an intention to treat analysis.
Per protocol analysis excludes patients who deviated from the protocol. It can introduce a form of bias called attrition bias, in which the groups of patients being compared no longer have similar characteristics.