This is something that the SMART initiative is working on. There are problems with two-stage cluster sampled surveys that may need addressing. Also, the "LQAS" method is being evaluated (it has a "mixed press"). Something needs to be done about surveying in nomadic populations (ACF is working on this and looking for pilot sites). FANTA has recently published work on a new mortality estimation method that shows promise (this is the subject of another thread on this forum). For me, the next big innovation is a move to using MUAC based case-definitions for prevalence estimation.

Thanks for the information. Eli