In my years as an editor of Anesthesiology and a reviewer for numerous journals, I frequently encountered submissions in which figures contained fitted curves (linear regression or other nonlinear functions) extrapolated beyond the range of measured X
values. Authors were often resistant to changing these figures to limit their predictions to the observed range of X
values. The folly of extrapolating beyond the observed range is seen in figure 4 in a recent review article in Anesthesiology 1
. The fitted function indicates that a decrease in blood volume increases cardiac output, certainly an unexpected and unphysiologic finding (and not supported by the data). Fortunately, the authors limited their extrapolation; had they extended their curve further, it would suggest that profound hypovolemia was desirable, certainly not the authors’ intention.
Dennis M. Fisher, M.D.