Sir,
I read this paper with interest the paper and I must confess to puzzlement
over the results of statistical analysis presented in Table I.
Three of
the four characteristics analysed are 2 x 2 contingency tables and the
remaining one (OTA classification distribution) a 3 x 2 table. The authors
state that they used unpaired Student t-tests and Fisher's exact test for
comparisons between the groups, the latter being the appropriate test for
analysis of such tables. Fisher's exact test produces one p-value for each
contingency table analysed, however, Table I in the paper reports a p-value
for each row. Furthermore, the quoted highly significant p-values for Type
B and C (both p < 0.001) seem unlikely when the percentages (76.8%
versus 70.0% and 22.2% versus 18.9%, respectively) are similar.
I would be grateful if the authors could explain these
inconsistencies.
J. Cooper BSc FRCS(Tr&Orth),
Consultant Orthopaedic Trauma Surgeon,
Selly Oak Hospital,
Birmingham, UK.