Lecture 1: Clinical trials Flashcards
What does a good clinical trial involve?
Comparison
new treatment vs standard treatment over time
What is the definition of a clinical trial?
Any form of planned experiment which involves patients and is designed to elucidate the most approptriate method of treatment for future patients with a given medical condition
What are 2 asepcts a clinical trial must provide?
Efficacy: the ability of a health care intervention to improve the health of a defined group under specific conditions
Safety: the ability of a health care intervention not to harm a defined group under specific conditions
What are the different phases of a clinical trial?
Phase 1: volunteer studies, <100 volunteers (looking at pharmacodynamics/pharmacokinetics/major side effects)
Phase 2: treatment studies, <1000 patients (looking at dosage and common side effects)
Phase 3: clinical trials, <10,000 patients (comparison with standard treatments)
Phase 4: post-marketing surveillance, whole population (monitoring for adverse reactions- yellow card reporting, potential new uses)
How do you get a fair comparison?
-ensure the 2 groups don’t know whether they are receiving the placebo/old drug or the new drug
How do you determine who should/shouldn’t take part in a clinical trial?
- inclusion criteria
- exclusion criteria
What was the inclusion criteria for the AstraZeneca vaccine clinical trial?
- ages 18-130
- all sexes
- healthy volunteers
- those at increased risk of SARS-CoV2 infection
What was the exclusion criteria for the AstraZeneca vaccine clinical trial?
- confirmed or susepcted immunosuppressive or immunodeficient state
- significant disease, disorder or finding
- prior or concomitant vaccine therapy for Covid-19
What are the reasons for pre-defining outcomes?
Need to define what, when and how outcomes are to be measured before the start of the clinical trial
- have a clear protocol for data collection
- agreed criteria for measurement and assessment of outcomes
- prevent ‘data-dredging’/’repeated analyses’
What are primary and secondary outcomes?
Primary
- preferably one primary outcome
- used in sample size calculation
Secondary
- other outcomes of interest
- often includes occurrence of side effects
What are the different types of outcome?
Patho-physiological e.g. tumour size/thyroxine levels/biomarkers
Clinically defined e.g. death (mortality), disease (morbidity), disability
Patient focused e.g. QoL, psychological well-being, social well-being, satisfaction
What are some features of an ideal outcome?
- appropriate and relevant
- valid and attritbutable
- sensitive and specific
- reiable and robust
- simple and sustainable
- cheap and timely
What are some timings of measurement?
- baseline measurement of relevant factors (monitor for inadvertent differences in groups)
- monitoring outcomes during the trial
- final measurement of outcomes (comparing final effect of treatments in trial)
How do we show comparability between groups?
- we need to try an ensure groups are compared equivalent as possible
- one was of demonstrating ‘comparability’ is by collecting baseline data on characteristics that we think may relate to both the condition and the outcomes we are investigating
What are some suggested baseline data for the covid vaccine?
-age
-gender
-ethnicity
-occupation exposure
-social class
-comorbidities
-BMI
To ensure the 2 groups are equivalent
What are the most important ethical considerations for any trial to go ahead?
- trials of new drugs may do harm
- patients/participants must understand what participation involves (including known and unknown risks)
What must clinical trials be to be able to give a fair comparison of effect and safety?
Reproducible- in experiemental conditions (record exactly how you did your experiment)
Controlled- comparison of interventions
Fair- unbiased without confounding (confounders are things that are related to the disease and the outcome e.g. alcohol/smoking in heart disease, so important that study has equal numbers of heavy drinkers/smokers in both arms of study)
What is the disadvantage of a non-randomised clinical trial and use of historical controls?
Comparison with historical control: e.g. looking at leic patients now with the drug compared to 2 years ago without the drug
-many other aspects that have changed in that period of time, not just the drug
Non-randomised: e.g. comparing leic patients who have access to the drug to norfolk patients who don’t have access to the drug
-differences within population e.g. ethnicities
-
What are non-randomised clinical trials and its problems?
Involve the allocation of patients receiving a new treatment to compare with a group of patients receiving the standard treatment
- confounding (known and unknown)
- allocation bias (allocating groups unfairly)
What is non-random allocation?
Allocation of participants to treatments by a person, historical basis, geographical location, convenience, numerical order etc.
-leads to potential for allocation bias and confounding factors to cause unidentified differences between the treatment groups being compared
What is comparison with historical controls and its problems?
Involves the comparison of a group of patients who had the standard treatment with a group of patients receiving the new treatment
-but for the standard treatment group: selection is often less well defined/less rigorous, unable to control for cofounders, less information about potential bias/cofounders, treated differently from the ‘new treatment’ group
Why is random allocation (randomisation) a good thing?
Minimal allocation bias: randomisation gives each participant an equal chance of being allocated to each of the treatments in the trial
Minimal confounding: in long run, randomisation leads to treatment groups that are likley to be similar in size and characteristics by chance
How do we do randomisation?
-toss a coin
-random number tables (even vs odd)
Most trials now use:
3rd party, computer generated random allocation
-randomisation of larger groups leads to less confounding
What is stratified randomisation?
When randomisation of a small group doesn’t work very well you can use this (you could just sample a larger group)
-stratifies the study into subgroups, with the same attributes, then followed by simple random sampling of the subgroups
What is the problem with knowing the treatment allocation (open label)?
Knowledge of which participant is receiving which treatment may bias the results of a clinical trial
e. g.
- patient may alter their behaviour, other treatment, or even expectation outcome (behaviour effect)
- clinician may alter their treatment, care and interest in the patient (non-treatment effect: not due to treatment)
- investigator may alter their approach when makng measurements and assessing outcomes (measurement bias)
(sometimes it has to be an open label trial)
Why do we blind clinical trials?
-remove allocation bias
What are the different types of blinding?
Single blind: one of patient/clinician/assessor does not know the treatment allocation (usually patient)
Double blind: two of patient/clinician/assessor does not know the treatment allocation (usually patient + clinician/assessor)
Triple blind (term rarely used as double blind usually implies all don’t know the allocation)
How do you establish blinding?
- aim to make treatments appear identical in every way (appearance, taste, texture, dosage regimen, warnings)
- use a designated pharmacy to label identical containers for the treatments with code numbers and have a code sheet detailing which code number responds to which treatment
What are some examples when blinding is difficult?
- surgical procedures
- psychotherapy vs anti-depressants
- alternative medicine e.g. acupuncture vs western medicine
- lifestyle interventions
- prevention programmes
What is confounding?
A factor that is associated with the disease and the outcome of interest, and this association is separate to the relationship between the risk factor being investigated
What is bias?
Systematic distortion in allocation/measurement
- may affect selection
- may effect outcome measurement by patient/dr
- publication bias
How do we remove confounding and bias?
Confounding: randomisation (should use a concealed allocation so no possibility of predicting next allocation of patient)
Bias: blinding (and randomisation)
What is the placebo effect?
Even if the therapy is irrelevant to the patients condition, the patients attitude to his/her illness may be improved by a feeling that something is being done about it
What is a placebo?
An inert substance made to appear indentical in every way to the active formulation with which it is being compared e.g. appearance, taste, texture, dosage regimen, warnings
Compare the active treatment and the placebo:
- the active treatment has the placebo effect PLUS the active treatment effect
- whereas the placebo only hasthe placebo effect
What are the ethical implications of placebo?
Use of a placebo is a form of deception
- a placebo should only be used when no standard treatment is available
- therefore it is essential that participants in a placebo-controlled trial are informed that they may receive a placebo
Why do you get losses to follow up?
Not every patient remains in the trial
- their clinical condition may necessitate their removal from the trial (appropriate)
- they may choose to withdraw from the trial (unfortunate)
How do you minimise losses to follow up?
- make the follow up practical and minimise inconvenience
- be honest about the commitment required from the participants
- maintain contact with the participants
- avoid coercion or inducements (payments for questionnaires etc, should be voluntary)
Why does non-compliance occur?
- they may have mis-understood the instructions
- may not like taking their treatment
- may think their treatment isn’t working
- may prefer to take another treatment
- they can’t be bothered to take their treatment
How do you maximise adherence with treatment?
-simplify the instructions
-ask about adherence
-ask about effects or side effects
-monitor adherence (tablet count/urine level/blood level)
(understand it is never possible to achieve 100% adherence)
What is an explanatory trial?
A study where you are looking for the best possible effect under the most perfect conditions
- efficacy study (produce a desired result under ideal conditions)
- do an ‘as treated’ analysis (exclude people who haven’t taken the treatment )= this affects randomisation
- non-compliers are likely to be systematically different from compliers= selection bias and confounding
What is a pragmatic/effectiveness study?
Looking for real world results
-intention to treat analysis
-take into account non-compliance
=compares the likely effects of using the treatments in routine clinical practice
-preserves the effects of randomisation and minimises selection bias and confounding
How does ‘as-treated’ differ from ‘intention to treat’?
As treated: analyses tend to give larger sizes of effect
Intention to treat: analyses tend to give smaller and more realistic sizes of effect
-clinical trials should be run on an intention to treat analysis
Example illustrating the two types of analysis:
New treatment Compliers: 70%, -40 mmHg Non-compliers: 30%, +5 mmHg Standard treatment Compliers: 100%, -30 mmHg
If you just look at compliers (as-treated), the new treatment appears better, but if you consider the non-compliers, the standard treatment appears better (intention to treat)
What are some considerations in an RCT (randomised controlled trial)?
- disease of interest
- treatments to be compared
- outcomes to be measured
- possible bias and confounders
- patients eligible for the trial
- patients to be excluded from the trial
How would you conduct the trial in an RCT?
- identify a source of eligible patients
- invite eligible patients to be in the trial
- consent patients willing to be in the trial
- allocate participants to the treatments fairly
- follow up participants in identical ways
- minimise losses to follow-up
- maximise compliance with treatments
How do you compare outcomes in an RCT?
- is there an observed difference between the treatments groups
- could the observed difference have arisen by chance? (is it statistically significant)
- how big is the observed difference between the two groups? (is it clinically important?)
- is the observed difference attributable to the treatments compared in the trial?
