Riffenburgh R.H. Statistics in Medicine

Academic Press. — 2005. — 672 p. 2 edition
ISBN: 0120887703To refer to the last half-century as revolutionary in the statistical world would
be a gross understatement. I believe the effects of this revolution are more profound
in the world of medicine and other health-related sciences than in most
scientific disciplines. Most studies in health sciences today involve formal comparisons,
requiring the use of appropriate statistical input on study design and
analysis. Some have become very sophisticated, as the FDA requires complex
and often large clinical trials, and most epidemiologic studies are more elaborate
than heretofore.
As a consequence, for those who aspire to engage in research activities, formal
training in statistics is a sine qua non. This is true even though most medical
research today is, by and large, collaborative, involving a team of scientists,
including at least one trained statistician. As with any endeavor, communication
among team members is crucial to success and requires a common language
to bridge the gaps between their disciplines. Thus, although it is certainly not
necessary or expedient for most research scientists to become statisticians or for
most statisticians to become medical doctors, it is essential that they develop
these avenues of communication.
In fact, all health care practitioners who even occasionally glance at their
voluminous literature encounter statistical jargon, concepts, notation, and reasoning.
Of necessity, therefore, the training of most such practitioners involves
some exposure to statistics in order that clear communication and understanding
result.
Those statisticians who collaborate with medical researchers struggle mightily
with this problem of communicating with their colleagues. Often this involves
writing texts or reference books designed not to turn medical researchers into
statisticians but rather to train them in a common language in order to be good
consumers of