The confidence attribute is only meaningful as part of
data, e.g. in the case of differential diagnosis, or else it may
be required on a fine-grained basis for data structures which have
varying measurement confidences, e.g. the example of a BP where
the diastolic sound could not be heard well. This latter situation
seems unlikely, but if true, would suggest that a confidence:REAL
attribute should be added to ITEM. Else it disappears altogether
to be taken care of by archetypes.