155 / 324
Statistical analysis was performed by the
Biostatistical Centre of the School of Public
Health of the Catholic University Leuven,
using the SAS statistical package version 9.1.
Multiple regression models were used in order
to examine gender and morphotype as predic-
tors for the measured femoral and tibial mor-
phologic dimensions. Likelihood-ratio tests
were used to compare the performance of
various nested models. Chi square tests were
used for comparing binary variables, two
sample t-test and Wilcoxon tests for compa-
ring independent variables in two groups (gen-
der). ANOVA and Kruskall-Wallis tests were
used for comparing variables in three groups
(morphotypes). Spearman correlation coeffi-
cients were used when appropriate The signifi-
cance level was set at p<0.01.
RESULTS
Distal Femoral geometry
Of the 250 smallest knees, 244 (98%) were
female, whereas of the 250 largest knees 203
(81%) were male. Of the 500 patients with
intermediate size knees, 104 (21%) were male
and 396 (79%) were female.
Smaller knees were significantly wider in
mediolateral versus anteroposterior ratio’s
than larger knees, both for males and females
(p<0.01) (fig. 3).
Female knees were significantly more narrow
than male knees (p<0.01).
14
es
JOURNÉES LYONNAISES DE CHIRURGIE DU GENOU
154
Fig. 2 : Morphotype characterization was based upon the pelvis width/total leg length ratio
as endomorph (left), mesomorph (middle), or ectomorph (right).