A multi-center, double-blind, placebo controlled study of paroxetine and imipramine in adolescents with unipolar major depression - efficacy and adverse outcomes.

Our statistical analysis plan is identical to that proposed in PROTOCOL NUMBER 29060/329 dated 12 June 1993

Criteria for Efficacy
1.1 Primary efficacy variables
a)   The change in total HAMD score from beginning of the treatment phase to the endpoint of the acute phase.
b)   The proportion of responders at the end of the eight week acute treatment phase.
1.2 Secondary efficacy variables
a)   Changes from baseline to endpoint in the following parameters:
•   Depression items in K -SAD-P
•   Global Impressions
•   Autonomic Function Checklist
•   Self Perception Profile
•   Sickness Impact Scale.
b)   Predictors of response (endogenous subtypes, age, prior episodes. duration and severity of present episode. comorbidity with separate anxiety, attention deficit. and conduct disorder).
c)   The number of patients who relapse during the maintenance phase.
2 Statistical Methods
2.1 Comparisons of interest
The comparison of primary interest is active treatment versus placebo. Hypotheses concerning these comparison will be tested at the alpha level of 0.05
2.2 Sample size determination
This study is designed to have adequate power to detect a clinically meaningful difference in both active-placebo comparisons at a two tailed alpha level of 0.05 and power 0.80. The sample size estimates are further based on an effect size of 0.40.
The rationale tor this effect size is as follows:
•   A difference of 4 in the HAMD Total change from baseline scores at endpoint. This is a smaller difference than that seen in previous studies with antidepressants in adults, yet it is large enough to be clinically meaningful, and
•   A standard deviation of 10. This is 20% larger than observed in studies with anti-depressants in adults and should reflect the greater variability in response expected in adolescent depression.
These parameter estimates result in 100 patients per treatment group.
3 Efficacy Analysis
3.1 Intent to Treat Analysis
All patients who receive double-blind medication will be considered as part of the ITT population. This patient population will be considered the primary population.
3.2 Patients Valid For The Efficacy Analysis
All patients randomized to study treatment and for whom at least one valid post-treatment efficacy evaluation is available will be valid for inclusion in an 'intent-to-treat' analysis. Patients who meet the following criteria will be eligible for the efficacy analysis:
a)   No major protocol violation exists with regard to inclusion or exclusion criteria
b)   No other major protocol violation during the first 8 weeks at active treatment has occurred.
Only primary efficacy variables will be analyzed using this population. Patients to be excluded from the efficacy analysis will be identified before the randomization code is broken.
3.3 Statistical Methodology
Psychometric scales using at least an ordinal measurement scale will be analyzed using parametric analysis of variance. Effects in the model will include treatment investigator and treatment by Investigator interaction. If the treatment by investigator interaction is not significant (p > 0.1) the interaction term will be dropped from the model. This analysis will be performed using the General Linear Models processing of the SAS system. The ordinal scales which have very few levels (such as the Severity of Illness) will also be analyzed using non parametric methodology so that the results are consistent across modes of analysis.
Dichotomous variables such as response (based on HAMO criteria) will be analyzed using Logistic Regression methodology. Effects in the model will include treatment, investigator, and treatment by investigator interaction; if the interaction is not significant then it will be dropped from the model. These analyses will be performed using the LOGISTIC procedure of the SAS system.
Summary statistics will be presented for demography, disease history, and baseline measures of efficacy.
An analysis of covariance will be performed to evaluate the effect of possibly important prognostic variables on the HAMD total score at endpoint. These include endogenous subtype, age at onset, gender, number of prior episodes, duration and severity of current episode, comorbidity with separate anxiety disorder, attention deficit disorder and conduct disorder.
3.4 Test of Significance
Tests of hypothesis regarding model assumptions such as the significance of treatment by investigator interactions will be made at the 10% level.
All other statistical tests will be two-tailed and performed at the 5% significance level.
3.5 Patient Characteristics At Baseline
Demographic and diagnostic variables at baseline will be checked for homogeneity between the treatment groups. If major differences exist for variables predictive of treatment response, their impact on the trial results will be investigated.
4 Safety Analysis
4.1 patients Valid for Clinical Safety & Tolerability
All patients who receive coded medication will be assessed for clinical safety and tolerability.
4.2 Adverse Experiences
Adverse experiences will be coded for each subject with reference to body system and preferred terms. The treatment groups will be compared regarding the incidence at the reported adverse experiences with reference to both preferred term and body system. The comparison between treatments with regard to incidence of adverse experiences will be performed primarily by using descriptive statistics.
4.3 Other Clinical Safety Variables
Information regarding demographic data, vital signs, physical examination, adverse experiences and abnormal laboratory values will be presented as listings and tables. All deviations from the study protocol and study withdrawals will be documented.