Clinical Trial Design Flashcards Preview

PoD > Clinical Trial Design > Flashcards

Flashcards in Clinical Trial Design Deck (14):

Why do a clinical trial?

Provides evidence, medical practice is evidence based but you you need to be able to critically evaluate evidence for what you do.

Use SIGN guidelines or Grampian Joint Formulary


Preclinical development

- Animal pharmacology (dose, adverse effects)
- Animal toxicology (teratogenicity, fertility, mutagenicity)
- Tissue culture


Phase I

- Volunteer studies
- Clinical pharmacology in normal volunteers generating pharmacokinetic, metabolic and pharmacodynamic data
- Usually involves around 100 subjects
- Certain drugs such as cytotoxics can bypass this phase


Phase II

- Clinical investigation to confirm kinetic and dynamics in patients
- Provides some evidence of efficacy and identifies a likely dosage range
- Involve up to 500 patients
- Looks at how the drug behaves in patients that may have comorbidities that affect how the drug works
- In case of cytotoxics it could be patients with late stage cancer and nothing to lose


Phase III

- Formal therapeutic trials where efficacy will be established and evidence safely obtained
- Involves 1000-3000 patients
- All data submitted on completion


Phase IV

- Post-marketing surveillance to produce evidence of long term safety
- May have involved tens or hundred of thousands of patients worldwide
- Clinicians have to report on first 10,000 pts
- There may be secondary benefits on another group of diseases


Types of trials

- Double blind: neither patient nor doctor knows
- Single blind: patient doesn't know, doctor does
- Prospective: more robust
- Retrospective


Placebo controlled study

- 50 patients on active drug and 50 patients on placebo: compare outcome.
- Then compare with other therapy so 50 on study drug and 50 on comparison
- Can then assess what is more effective


Cross over design

- 50 patients on study drug and 50 on compared
- Cross over after a set amount of time and a wash out period
- Then the patients are swapped



- Gold standard
- Patients assigned at random to either treatments or control


Disadvantages of randomised

Subjects may not represent general patient population
Tend to be better at complying with medication
Need twice as many patients
Administratively complex
Some physicians and patients refuse


Designing a study

- Need definitive hypothesis and end point
- Superiority or non-inferiority? Better than control or not worse?
- Choice of placebo or control drug... 50% placebo effective for anxiety
- Choice of patients: age and sex matched, race, other disease and drugs
- Are pts going to drop out? Comply?


Importance of statistical power

- Need to choose statistical test
- Analyse if differences are due to chance...
- p<0.05 is usually taken as significance


How can something have statistical significance and lack clinical significance?

- A drug may prove to be statistically good but not any better than the one on market at the moment
- Or it might be that there's evidence that there's no long term effect e.g. statins and cholesterol