Natural ideas, discontinuities between

These two extreme attitudes might be regarded as leading naturally to intention to treat in the first case and per-protocol in the second. These analyses are not, however, the only ones possible and there has been a great deal of attention to alternative approaches whereby the complex interactions between outcomes on the one hand and discontinuation and degree of compliance on the other, as well as the effect of treatment on both, are all modelled. The idea behind this is to recover, to an appropriate degree, the information on the effect of treatment given by noncompliers. 

---