Scientific method of detecting differences between treatments
Detect merits of specific treatments for patients with specific diseases
Provide evidence of efficacy and safety

Ethics behind clinical trials (declaration of Helsinki 1961)

1. Patients should not be denied effective treatments
2. New treatment must be safe (no patient should suffer as part of a trial) 

Pre clinical trials

• Assess toxicity / pharmacology of drug
• In vitro / in vivo testing

Clinical Trials

• Phase I: Normal healthy volunteers - to assess correct pharmacological dosing, route of administration (<20 patients)
• Phase II: Select subpopulation of patients, establish efficacy; resource assessment, if not helpful would not be ethical to proceed (-100 patients)
• Phase III: "Normal patients" (1000s) - establish efficacy and safety
• Phase IV: Pos-marketing surveillance

Controlled clinical trial

• Active treatment compared with control treatment (may be placebo, current standard, etc)
• Used to determine the best course of treatment

Clinical trial protocol

1. Introduction
2. Aims + hypothesis / precise question asked
3. Materials
4. Methods: end points
5. Results
6. Statistics
7. Bibliography
8. + Financial support, responsibilities of workers, signatures

Error / null hypothesis

Null hypothesis states that there is no difference between two treatment groups

1. Type I:
   • null hypothesis rejected despite being true
   • detecting a difference when one does not really exist
2. Type II:
   • null hypothesis accepted when it is false
   • Failure to detect a difference when one actually does exist

Power

• Ability of trial to detect an actual difference
• Equal to type II error

Significance level

• Statistical probability of a type I error

Confidence interval

• Probability of a true population mean lying within a range derived from a sample mean and it's standard error (standard error = standard deviation/number of observations)

Sensitivity / Specificity

1. Sensitivity: ability to identify a true positive
2. Specificity: ability to exclude a false negative

Averages / Measures of spread

• Mean, median, mode
• Standard deviation: - measure of scatter around the mean

Clinical trial designs

• Randomised (minises bias)
• Case - control
• Cross over
• Double-blind (useful when trial has subjective endpoints)