Coarse Grain,Fine Grain distinctions in Medicine.

Jacob Bronowski in his "The Origins of Knowledge and Imagination" speaks of a basic problem in the brain's mechanism in terms of achieving fine discrimination with a coarse apparatus.Further, he says that in many ways "about all" human problems..in science or in literature..center around the same problem.How do you refine the detail with an apparatus which remains at bottom coarse and grainy?"
Consider the randomized clinical trial(RCT) currently thought to be at the apex of the mechanisms we have to find the medical truth.We are left with aggregate data, groups numbers that are relevant , strictly speaking, to patients having the characteristics of those in the trial but is freqently extrapolated to include many other patients.
Compare this grainy-ness with the progress made in the following example.
Consider the use of MRI in a patient with the clinical picture of optic neuritis.
The finding of white matter abnormalities of a cetain type provide valuable information regarding likelihood of progression to multiple sclerosis. The coarse grain category of optic neuritis has been fine grained a bit by sorting out those patients with a high risk of subsequent development of multiple sclerosis.
But fine tuning cannot occur with the RCT. We can break the aggregate data into sub groups but we loose power to see differences and at the same time we increase the likelihood of Type I errors by cranking out many comparisons. So basically the RCT is a coarse grain apparatus and cannot get any finer, remaining coarse and grainy.
To fine tune the aggregate data seems to take determining the relevant pathophysiology.The example of TPMT deficiency comes to mind. Before the details of this deficiency were mapped out, we could only say that a certain percentage of children treated with a thiopurine type drug developed serious bone marrow failure. Once we learned what the pathophysiology was , testing could be done to see who comprised that percentage and the coarse grain became fine.