Clinical Trial Design Flashcards
why are clinical trials important
they provide evidence which influences medical practice
4 examples of drug treatments based on trial evidence
MI
stroke
cancers
rheumatoid arthritis
investigating potential drugs (5)
- does it work
- what dose is therapeutic
- what dose is toxic
- is it safe
- is it necessary
problems with observational studies (3)
correlation vs causation false + ves
replication is difficult
why conduct robust clinical trials
what works in theory might not be best in practice
e.g. high [oxygen] in premature babies - found to be harmful tonsillectomy
-generally unnecessary bypass surgery vs coronary artery stenting
Clinical trials are regulated by
MHRA- tests efficacy and safety
stages in development (4)
drug discovery
preclinical development
clinical development
(volunteer studies, phase i)phase ii, phase iii, phase iv
pre clinical development (3)
• Animal pharmacology (dose, adverse effects)
• Animal toxicology (teratogenicity, fertility, mutagenicity)- Tissue culture
clinical development: phase i
• Clinical pharmacology in normal volunteers generating pharmacokinetic, metabolic and pharmacodynamic data
• Usually involves around 100 subjects
• Certain drugs e.g. cytotoxics will bypass this phase e.g.
• TEGENERO DRUG: 8 paid volunteers
○ 6 given active drug IV, 2 given placebo
○ Regulated environment, according to protocol approved by MHRA
Developed multiple organ failure in an unpredicted biological action of the drug in humans
phase ii
• Clinical investigation to confirm kinetics and dynamics in patients (who may have liver/renal/GI absorption problems)
• Provides some evidence of efficacy and identifies a likely dosage rangeInvolves up to 500 patients
phase iii
• Formal therapeutic trials where efficacy will be established and evidence of safety obtained i.e. does it work for the condition we are testing
• Involves 1000 - 3000 patientsAt completion, all data (pre-clinical, pharmaceutical and clinical data) is submitted as an application to the regulatory authority for a license to sell the drug
phase iv
• Post-marketing surveillance to produce evidence of long term safety May involve tens or hundreds of thousands of patients world wide
clinical trials
pilot studies
double blind/single blind
retrospective/prospective
pilot studies
not to estimate outcome but to test study design
double blind
patient and doctor blinded
single blind
patient blinded
prospective
protocol decided before hand
retrospective
less good as open to bias
placebo controlled study
100 patients
50 drug
50 placebo
compare outcome in the 2 groups
comparison with other therapy
100 patients
50 study drug
50 comparative therapy
compare end pointsis one better than the other
cross over design
100 patients
50 study drug
50
comparative therapy cross over at e.g. 8 wks (wash out period)compare outcomes A vs B in same patient
randomised clinical trial
patients assigned at random to either treatment or control this is the ideal method
disadvantages of randomised control clinical trials (4)
- Generalisable results?
• Subjects may not represent general patient population
• Tend to be better at complying - Recruitment
• Twice as many new patients needed for the study - Acceptability of randomisation process
• Some physicians will refuse (PFO closure)
• Some patients will refuse (want treatment)
Administrative complexity (randomisation methods etc)
commonly used phase iii designs (9)
- Parallel
- Withdrawal
- Group/cluster
- Randomised consent
- Cross over
- Factorial
- Large simple
- Equivalence/non-inferiority Sequential
superiority vs non-inferiority trials
- SUPERIORITY: show that new treatment is better than the control or standard (maybe a placebo)
- NON-INFERIORITY: show that the new treatment:
A. Is’nt worse than the standard by more than some margin
B. Would have beaten placebo is a placebo arm had been included (regulatory)
how to design a study
- Randomised double blind comparison of angiotensin II antagonist vs atenolol in the treatment of hypertension
- 12 wk comparison
- End points (should be as simple as possible)
• Death
• Number of hospital admissions • Lowering of blood pressure
• Compare with pain control or change in mood - Hypothesis
- Number of subjects
- Safety endpoints? - A good design will give robust results
designing a study
- Choice of subjects: need enough to be able to detect/reject a difference between the 2 groups, statistical design is very important
- Number of patients also depends on: frequency of outcome measurement (A v B in mild hypertensives, BP reduction: 200 patients over 12 wks; stroke reduction: thousands of patients over 5 yrs)
- Choice of control drug: placebo, drug of known efficacy
- Choice of patients: age and sex matched, race, other diseases and drugs, compliance
- Exclusion and selection criteria: exclude pregnant women, children, seriously ill patients, elderly (?), patients at risk of side effects
challenges in studies (2)
- Declining renal function Multiple morbidities: patients over 60 usually have multiple conditions
analysis and interpretation
- Choose a stat test - Are differences due to chance
- P<0.05 usually taken as significance- Interpreting an insignificant finding:
• No difference or the study hasn’t found one Two treatments may be clinically effective
ethical issues
- Consent
- Ethics committee
- Placebos
- Children
- Study design
- Policing studies
- MHRA/CSM/EU
- InsuranceThe law
all preclinical and clinical trial evidence is submitted to the …
regulatory authority
post marketing surveillance
- Medicines and healthcare devices regulatory authority (MHRA)
• Committee of safety of medicines- Yellow card systemGP, hospital doctors and pharmacists
why do we do clinical trials
- Patients may get better or worse despite drug therapy
- There may be a placebo effect
- Clinical practice needs to be evidence basedDrug companies have competing interests
important clinical trials
4Scare trial
scandinavian simvastatin survival trial
4444 patients with angina or post MI, serum cholesterol 5.5 -8.0mmol/L, simvastatin 20mg or placebodeath lower 4%, coronary events lower 9% in simvastatin
cholesterol and recurrent events trial
does lowering normal cholesterol give benefit, 4000 patients, post MI, ‘normal cholesterol’, pravastatin 40mg or placeboStroke incidence reduced by 31%, p = 0.03
