Statistical analysis was conducted using INSTATTM
for Windows. Our primary outcome was clinical
improvement, which was defined as a reduction of
20 mm on the VAS for pain. Clinically significant
effect size was determined using an odds ratio
(OR) and number needed to treat (NNT) with
32 H. Gemmell et al.
95% confidence intervals. The difference in posttest
VAS, cervical lateral flexion and pain threshold
scores between the groups was determined using
one-way analysis of variance with alpha set at
0.05.