Statistics of embryonic stem cell research?

I have been doing some reading of the various journal articles on esc research and have discovered that a lot of times their statistical ysis is flawed (I'm a mathematician, not a clinical researcher). For example, if an experiment studied 20 rats and 8 of them died, most of the time the result of those 8 rats are not included in the final ysis, showing a much higher rate of "cure" than there actualy was. Has anyone else noticed this? Has anyone done a literature review to see if this is prevailant in the science literature? Could I just have had too small of a sample of the literature?