Clinical Trial Methodology Flashcards
What is meant by a clinical trial
A clinical trial is a planned experiment in humans, designed to measure the effectiveness of an intervention. The intervention is usually a new drug, but the method can equally be applied to the assessment of a surgical procedure, a vaccine, complementary therapy etc.
Describe how clinical trials differ from other types of studies
Experimental studies like trials are different from most epidemiological studies (surveys, cross sectional, cohort, case control, ecological) which are observational. In observational studies the investigator measures what happens but does not control it. For example, an investigator may record whether people smoke, and relates this to whether or not they develop lung cancer. In contrast, in a clinical trial, the investigator would allocate one group to smoking and the others to not smoking, and then see who got ill.
What happens in an ecological study
Summary estimate of two factors, compare, give starting point. Summary estimates of two factors- make inferences (useful when the data is hard to measure)
Describe the basic plan for a RCT
Planning an intervention (e.g. drug or surgery)
Define the population
o Defining the comparator (e.g. placebo, alternate treatment, standard of care).
o Defining the inclusion/exclusion criteria.
o Defining the control group – people eligible for the intervention but receive the comparator.
What do you do when looking at population based studies
Infer and observe potential causal relationships.
What are the uses of clinical trials
Is the drug better than no drug- used for assessing safety
Is the drug better than the best existing drug
What/how common are S/E?
‘Proof’ of causation?
What are the critical features in the design of a clinical trial
Objectives Patients selection Controls Study size Unbiased data collection Specific design Ethics Analysis
What do we need to define in a clinical trial
- entry criteria
- disease etc
- measures of response
- end points
How can we eliminate chance
Large sample size, larger number of events- this increases the power of the study.
What is meant by bias
A systematic error in design, conduct or analysis of a study which produces a mistaken estimate of an exposure on the risk of disease
How can we eliminate bias
Design of study- randomisation- this also eliminates confounding. Analyse whether this was done effectively when critically appraising articles.
What should you look for when assessing bias
- randomisation
- after entry- randomisation should happen before entry into the study
- investigator blind
What is particularly important to do when carrying out trials in different geographical locations
Stratify the patients into different groups, to ensure that the outcome can be generalised to all populations.
What are the advantages of randomisation
Validates statistics Excludes biased allocation Equally distributes prognostic factors? known unknown - need stratification? - restricted/block randomisation
What are the methods of minimising bias
o Blinding – single (patient not knowing), double (clinical team not knowing) and triple (statistician doesn’t know).
o Accurate end-point selection.
o Minimising loss to follow-up.
Why do we need to randomise
People who are eligible for the trial (i.e. have the condition you are interested in) are recruited, consent is obtained and then they are randomly allocated to the intervention or control
groups. Randomisation is done to remove treatment allocation bias. Without randomisation it is possible (indeed likely) that the investigator will choose different patients for each group. In
a famous early study of the BGC vaccine for TB in children, deaths from TB were five times higher in the control group than the vaccinated children.1 Further investigation showed that doctors had tended to offer the new vaccine to children whose parents were more ”cooperative”, and left the rest as controls.
These cooperative parents were likely to have been more educated, health conscious and therefore to have a lower mortality from TB regardless of the vaccination. Hence the need for randomisation in treatment allocation.
What is the importance of a control group
The control group is those study participants who do not receive the intervention under assessment. A control group must be included otherwise you cannot be sure why the outcome happened; it may be due to the new treatment or it may have happened anyway. Control groups may be given a placebo (an inactive substance such as a sugar pill, or water injection), or a standard treatment Placebo effect Regression to the mean Acclimatisation Seasonal effect
Describe regression to the mean
Any intervention at bad times will be followed by improvement, was this due to the intervention or not.
Describe controls
- are the basis of the comparison
Without them there is no reason to assume that the observed effect is due to the intervention.
What do we need to know about the sample size
Objective?
Difference to be detected (minimum important)
Significance of obs. difference = (p)
Confidence by which a-ve results is genuine = ß (1-power)
Variation of data (SD)
Are sponsors involved in conduct of the study
No, this is done by a separate academic committee.
Describe the potential sources of bias in RCTs
Sponsorship Study patient selection/allocation Prejudice of patient Prejudice of observer Faulty method Faulty analyses Faulty interpretation- any point in continuing when the results are not answering your hypothesis? If patients are dying as a result of the intervention, you need to stop?
Describe the relationship between sponsorships and RCTs
Critical requirement Data ownership Design issues Independent analyses Independent reporting
Which trials are sponsors more likely to fund
low relative risk trials
What are the pros of blinding
Doctors knowledge bias - gone
Patient’s knowledge bias - gone
Withdrawal bias - gone
What are the cons of blinding
Impossible?
Ethics?
‘Cost’ – complex
Titration
Why do we blind
Blinding means that the patient does not know whether they are getting the new treatment or not. In a double blind trial neither the patient nor the doctor knows which treatment they are getting. This is to prevent bias in reporting or measurement of the outcome, measurement bias. People who are getting a new treatment (or treatment compared with no treatment) often report improvement in subjective symptoms because they are enthusiastic and hopeful. Similarly if a doctor knows that a patient is on the new or active drug they may look for more improvements.
What is a consequence of analysing data from sub-groups
Less end events, hence less power, more likely to be due to chance rather than a true association.
Describe cross-over studies
Short-term, small sample size, low budget trials
Each person can be their own controls
Switch groups around (include a wash-out period to prevent spill over of effects)
Observe again
This increases the power of the results.
What are the potential problems when interpreting a trial
‘Healthy volunteer’ effect Baseline BP level - mean diastolic: 98 or 91mmHg? 3. Active treatment of placebo group Drugs: Composition and dosage
Describe a parallel study
Big numbers Curative Long R effect Surgery Carry over effect Long duration Parallel, where individuals are assigned to different interventions (usually 2, the intervention and control arm, but there may be >2 arms).
Describe the ethical aspects that we need to consider
No unnecessary risk Benefit? Placebo? Committee approval Informed consent Entry criteria Withdrawal criteria Stop the study criteria
Why are ethics and consent important
Clinical trials are strictly regulated to ensure that patients are protected. All clinical trials have to be registered, reviewed by an independent scientific committee, be approved by a Research Ethics Committee and adhere to government and international guidelines. Trials will have an independent data monitoring committee – a group of independent researchers who can check progress during the trial; they will usually unblind the results to see if there is any major difference in outcome (improvement or side effects) between the intervention and control groups. If there is a large difference they have the power to stop the trial.
All participants in a trial must provide informed consent, and be free to withdraw at any time without affecting their care.
Describe the intention to treat or per protocol analysis
In many trials there will be some deviation from the trial design. For example, some people who are randomized to a treatment will stop taking it because of side effects. Others will shift to another treatment for clinical reasons. A decision then has to be made about what to do with people who deviate from protocol. This depends on the aim of the study. Usually the trial investigators should report intention to treat analysis as only this maintains the original randomization and therefore differences between the groups should reflect the treatment and not other (confounding) factors. If the aim is to assess efficacy of the treatment then it may be appropriate to analyse those who stuck to the protocol. However, this will give less idea of how the treatment will work in practice (effectiveness). If, for example, one-third of people stop taking the treatment because of side effects then this will affect the effectiveness of a strategy that makes it a first-line treatment
Describe open-label studies
Sometimes it is not possible to conduct a blind or double-blind trial, e.g. when a surgical intervention is compared with a medical treatment. However, it is still important to include randomization and controls and, where possible, to ensure that people doing outcome measurements are blinded to which treatment the participant received.
Describe cluster studies
• Cluster trials, where groups of patients, clinics, or even geographic areas form the basis of the intervention, and the outcome is compared between these groups rather than between individuals. This is appropriate if an intervention is likely to have a group or population effect (e.g. testing a vaccine or other preventive intervention) or if interventions aimed at some people are likely to spill over to others in the community, e.g. a dietary trial.
