changes in symptom severity between groups in
relation to the pooled standard deviation.30 Effect
sizes were categorized as small (0.2 to 0.4),
medium (0.5 to 0.7), or large (0.8 or greater). For
example, in a group of patients with treatmentresistant
schizophrenia, switching from treatment
with typical antipsychotic agents to optimal clozapine
treatment was associated with a large effect
size (0.8) for improvement of overall symptom
severity and a medium effect size (0.5) for
improvement of specific positive and negative
symptoms.31 The relative benefit of atypical as
compared with typical antipsychotic agents for
long-term memory dysfunction in schizophrenia
was associated with a small effect size (0.2).32
In the present study, the primary analysis was
performed according to the intention-to-treat principle.
The primary outcome measure was the total
PANSS score. A mixed-model analysis was used
(SPSS, version 13), which included all available
data at each time point. This analysis included
fixed effects for group (risperidone and placebo)
and time (days 7, 35, and 63) and an unstructured
covariance structure. The statistic analyzed forsignificance was the interaction term, which allows
comparison of the treatment groups over
time. For descriptive purposes, we also provide
information on the differences between groups
in the change from baseline to eight weeks and
on effect sizes. In a secondary analysis, the proportions
of patients having a response (those with
a decrease in total PANSS score of at least 20
percent) in each group were compared by the
chi-square test.