3 Clinical psychopharmacology3.1 General pharmacology
3.1.5.2 Phase III the RCT in psychiatry
1 min read1 figure
Written by Resident Doctor Committee · Published
What a phase III trial does
Phase III trials test benefit and common harms in a defined population, with the population, intervention, comparator, outcomes and follow-up fixed before recruitment opens.
Allocation concealment hides the next allocation during recruitment, and masking limits differences in treatment and assessment afterwards. Neither replaces the other.
Outcomes in psychiatric trials
Outcome | What it measures | Limitation |
|---|---|---|
Continuous symptom change | Mean change on a rating scale | A statistical difference may not matter clinically |
Response | A prespecified reduction, often 50% | Dichotomising loses information |
Remission | Symptoms below a threshold | Function may stay impaired |
All-cause discontinuation | Stopping for any reason | Mixes benefit, harm and preference |
The CATIE schizophrenia trial
All-cause discontinuation was the primary outcome of phase 1.
About three quarters of patients (74% of 1,493) stopped their allocated drug within 18 months.
Olanzapine had the lowest stopping rate, still 64%, at a metabolic cost.
Perphenazine did about as well as most second-generation drugs.
Superiority and non-inferiority
Superiority needs the confidence interval to exclude no difference, and failing to show it does not establish equivalence.
Non-inferiority asks whether a new drug is no worse than the comparator by more than a prespecified margin, delta, so the lower confidence bound must exceed minus delta.
Reading the result
The standardised mean difference (effect size, Cohen's d) is the difference between group means divided by the pooled standard deviation.
It expresses the drug-placebo difference in units of the outcome's spread, so trials using different scales can be compared.
Relative effects need the baseline risk: halving 40% and halving 4% give very different absolute benefits.