Study Designs 2 - RCTs Flashcards
What is a clinical trial?
A scientific way of testing a clinical question by comparison with control group
What might you want to assess in a clinical trial?
Treatments Supportive care Devices Screening programmes Information Diagnostic tests Biomarkers
What key lessons have we learnt in history from clinical trials?
- James Lind & Scurvy: orange + lemons have the best effect when everything else was kept constant
- Gold therapy for TB: controls had better improvement
- Natural history of disease + vaccine therapy: no better than placebo
What have lessons from history taught us?
- Collect data i.e. dont rely on anecdotal evidence
- Need for controls
- Avoids bias - groups should be comparable except for treatment
What principles must be followed when comparing treatment groups?
- Groups alike in all important aspects except for treatments under evaluation getting rid of possible bias
- Large enough sample to limit chance imbalance
= difference in outcome is just due to treatment effect
Why use randomisation?
Best way to create a CONTROL group similar to the TREATMENT group in all respects (known and unknown)
What is a randomised clinical trial (RCT)?
Controlled clinical trial where therapies allocated by a chance mechanism where neither patient nor physician knows in advance which therapy will be assigned
What are the advantages of randomised clinical trials (RCTs)?
- Eliminate selection bias
- Balance prognostic factors
- Validity of statistical tests
What is a controlled clinical trial?
Prospective study comparing effects and value of an intervention against a control in human subjects
What is an uncontrolled clinical trial?
Involves NO control group
What is the main dictator of how good a study design is?
Reducing bias to an absolute minimum
In order of worst to best, what types of studies have the best study design?
- Uncontrolled
- Historical controls
- Current non-randomised controlled
- Randomised controlled
What is the concept of statistical interference?
Analysed data are results in a selected sample and we use these results to infer how all the population would behave - start with pop. and get protocol treated patients, look at observed results and extrapolate then:
- Estimate: treatment effect (beware bias)
- Beware of variability of estimate due to sampling from population
= Goal: control bias + reduce variability
What is the benefits of random allocation of treatment?
- Gives equal chance of receiving each treatment
- In long run leads to groups likely to be similar in characteristics by chance
- Reduces selection bias if patients enter trial before randomisation
- Used in other experimental settings
What is the placebo effect?
Even if therapy is irrelevant to the patients condition, the patients attitude to his or her illness, and indeed the illness itself, may be improved by a feeling that something is being done about it
A difference between results of treatment and no treatment group could be one of 2 things, what are these?
- True treatment effect
2. Placebo effect as one group is receiving care
What is the aim of blind trials?
Aims to remove differential placebo effect that could bias comparison between treatments i.e. strengthening randomisation
What is a single blind trial?
Where one of patient, clinician or assessor does not know the treatment allocation - usually PATIENT
What is a double blind trial?
2 or more of patient, clinician or assessor does not know the treatment allocation - usually PATIENT and CLINICIAN/ASSESSOR
Give some examples of blinding.
- Make treatment appear identical in taste, appearance, texture, dosage, regime etc.
- Compare active drug with placebo (inert substance identical in appearance, taste, texture etc.)3. Use a designated pharmacy to label identical containers with code numbers
How is a trial conducted?
- Eligible patient presents
- Consents
- Randomised to either new or control treatment
- Assessment
How are trial outcomes evaluated?
Compare the outcomes fairly to see:
- Is there a difference between groups?
- Could the difference have arisen by chance or is it statistically significant?
- How big is the difference, is it clinically important?
- Is the difference attributable to treatments in the trial?
Why might participants not remain in the trial?
- Clinical condition may necessitate removal from trial
- May chose to withdraw
Why might participants not comply with allocated treatment?
- Misunderstood instructions
- May not like taking treatment
- May think its not working
Why do we want to keep patients in trials?
So data is not skewed or bias i.e. if a patient is removed due to treatment affecting them negatively, if they are removed along with all their data this wont be recorded
How can we increase compliance?
- By giving medicines in chunks e.g. 2-weekly doses rather than 6-montly
- Give them incentives e.g. group exercises
- Make goals realistic e.g. for obesity and exercise
How are patients randomised?
Either to NEW treatment (either take new treatment or not) OR STANDARD treatment (take control treatment or not)
How can be determine if the new treatment is better than the standard?
2 different interpretations:
- Is the PHYSIOLOGICAL ACTION of the new treatment better than the standard?
- Is new treatment better than standard treatment IN ROUTINE CLINICAL PRACTICE?
What is an explanatory trial?
I.E. ‘as-treated’ analysis where you compare just the patients that have taken the new treatment with those that have taken the control treatment - those who completed follow-up and complied with treatments comparing physiological effects of treatment - tend to give larger sizes of effect
What is the disadvantage of an explanatory trial?
Loses effect of randomisation as non-compliers are likely to be systemically different from compliers so there is selection bias and confounding
What is a pragmatic trial?
I.E. ‘intention-to-treat’ analysis where both groups in the new and standard treatment groups are compared - tend to give smaller effect sizes and reflect effect in clinical practice
What type of analysis is best used for definitive clinical trials?
‘Intention-to-treat’ basis
What is the ethical dilemma of clinical trials?
Clinician should provide best treatment for each individual patient but scientific integrity requires treatment chosen randomly so can these requirements be reconciled?
What is clinical equipoise?
Where reasonable uncertainty about which treatment (including non-treatment) is better when putting a patient into trials so randomisation does not deny any patient the best treatment - the only qualification is ignorance
What should be explained to the patient when gaining informed consent for a trial?
- That the patient is invited to be in a trial
- What the alternative treatments are (including known side effects)
- That treatment will be allocated at random
- That patients may withdraw at any time
How should informed consent information be given?
Verbally and in writing with a ‘cooling off’ time for patients to think and reflect about the information (24hrs) by a knowledgeable informant
What process does standard randomisation use?
- Are patients eligible?
- If yes, patient consent
- If they dont consent, drop them from trial
- If they consent, randomise to treatment A or B
What process does the Zelen method use?
- Are patients eligible?
- If yes, randomise
- No consent is needed for standard treatment A
- Seek consent
- If consent is given, give treatment B
- If consent is not given, give standard treatment A
Why might the Zelen method be beneficial to use?
Because when giving a patient all information about a trial for informed consent, they may end up in the non-treatment group knowing that are missing out on treatment - this method avoids this as the non-treatment group do not need to be informed or give consent
