The elements of statistics for clinical readers
A. J. Hartz and R. L. Holloway
Department of Family and Community Medicine, Medical College of Wisconsin, Milwaukee, USA.
Statistical courses usually emphasize teaching the mathematical properties
of methods to perform statistical significance testing. To interpret
clinical studies, however, the reader needs to know only the purpose of the
statistical methods, not their mathematical basis. This purpose in most
studies is to evaluate the association between a risk factor and an
outcome. The evaluation has five components: (1) measure the strength of
the association, (2) determine the probability that the observed
association did not occur by chance alone, (3) find the range of probable
values for the measure of association, (4) reduce the possibility that the
association is invalid because of confounding factors, and (5) examine the
possibility that the association does not apply equally well to all people
because of modulating factors. By focusing on the purposes of the
statistical evaluation, the reader will be less distracted by the specific
mathematical formulations that provide little additional information to the
clinician.
