One problem I see in much of medical research is the difference between the statistical metrics of research and the actual clinical application. Unless the statistics are done very carefully, we tend to see anomalies as outliers, whereas in a clinical setting, that's where the problem is.
This can apply in various ways. For example, what if the metabolism differs between rats based on genetics or some other factor? By eliminating genetic diversity in a study, we might have chosen a case where an effect might express particularly strongly or weakly, or even in reverse.
And as another example, note that in a clinical setting, if a medication doesn't work, then an alternative is often tried. Those for whom a medication has a paradoxical effect (e.g., an antidepressant that depresses mood in some people) do not continue taking the medication and shouldn't bring down the "score" of efficacy--but in research, a medication that works for half of the people might be seen as having no effect if it's averaged in with those for whom it makes things worse.
I have come to realize that many doctors are not good scientists, and that the pressure for funding leads to overreaching on conclusions.
and wow, I wrote that very poorly. I have non-parallelism, misdirected references, etc. Not sure my points came across...
Bingo. The money has corrupted much of research. It's unfortunate but true.
Clinical trials would be great; but they're expensive. The anti-salt folks have been claiming since forever that salt increases blood pressure and that countless lives can be saved if we would just start dictating the amount of salt the food industry (foodservice included) can use. These same people run from the issue like roaches from light when the suggestion of randomized clinical trials to prove what they claim is suggested.
Tells you everything you need to know.