6
Why not sample size calculations?
■
A cynic once said that sample size calculations are a
guess masquerading as mathematics. To perform such a
calculation we often need information on factors such as
the standard deviation of the outcome which may not be
available. Moreover the calculations are quite sensitive
to some of these assumptions.
■
Any study, whatever the size, contributes information,
and therefore could be worthwhile and several small
studies, pooled together in a meta-analysis are more
generalisable than one big study.
■
Often, the size of studies is determined by practicalities,
such as the number of available patients, resources, time
and the level of finance available.
■
Studies, including clinical trials, often have several
outcomes, such as benefit and adverse events, each of
which will require a different sample size.