Randomised Controlled Studies Flashcards
What is a clinical trial
A clinical trial is an experiment in which a treatment is administered to humans in order to evaluate its efficacy and safety
What are the types of clinical trial
- An uncontrolled trial
- A controlled trial
- A randomised controlled trial
What happens in an uncontrolled trial
– Everyone gets the treatment – used that so all participants without the treatment in the trial will die in a few months
What happens in a controlled trial
– A treated group is compared with an untreated group (placebo)
– Or a treated group is compared with a control group having “usual treatment”
– Controls may be geographical, historical, or randomised;
How is allocation determined in a randomised trial
– Allocation to groups is determined by chance
What is a geographical control and what is it subjected to
- Patients with the same disorder seen at another hospital or clinic where the new intervention is not provided
- Selection bias – patients might be more likely to die in that area anyway not necessarily due to the treatment or lack of treatment
What is a hospital control and what is it subjected to
- Patients with the same disorder seen in the past before the use of the new intervention
- Similar problem as with geographical controls (selection bias)
- More biased to therapy being better
- Can be used if outcome of standard treatment (or no treatment) are well known and vary little for a given patient population
What are the two types of controls
Geographical control
Hospital control
What is an allocation bias
Allocation bias can be defined as bias that arises from a systematic difference in how participants are assigned to treatment groups and comparison groups in a clinical trial
e.g. allocation bias may result if investigators know or predict which intervention the next eligible participant is supposed to get
What is a selection bias
Selection bias is the bias introduced by the selection of individuals, groups or data for analysis in such a way that proper randomization is not achieved, thereby ensuring that the sample obtained is not representative of the population intended to be analyzed
e.g. Examples of sampling bias include self-selection, pre-screening of trial participants, discounting trial subjects/tests that did not run to completion and migration bias by excluding subjects who have recently moved into or out of the study area
Why can you not use alternative allocation in a randomized control study
You cannot do alternate allocation as the clinicians and patients can predict the treatment to be received therefore it is not random
How was the first trial randomised
Patients randomly allocated to receive streptomycin or bed rest
each centre allotted a numbered series of envelopes, each containing a card indicating ‘S’ or ‘C’. Numerical order of envelopes based on a series of random numbers. When patient approved for trial the next envelope was opened
What are the benefits of randomised control studies
- Proper randomisation helps ensure group receiving treatment A is similar to group receiving treatment B
- Avoids selection/allocation bias
- The only systematic difference between treatment and control groups is the treatment (hopefully (doesn’t guarantee it))
what do patients have need to be before randomised
Patients need to have been deemed eligible and meet the inclusion criteria and consented to participate for a randomised control study
What are the two different types of blinding
single blind
double blind
What is a single blind trial
Single blind – patients do not know what treatment they are on but researchers do
What is a double blind trial
Double blind – also the observers do not know what treatment the patients are on (not always possible)and the patients do not know what treatment they are on
What does blinding ensure
• Ensures use of other potential treatments/ assessment of outcome/decision to withdraw patient not influenced by clinician’s or patient’s knowledge of treatment
What are the two different types of randomised controlled trials
- Parallel groups
- Crossover
describe the parallel group type of the randomised controlled trials
randomise participants into group A and group B
- give them the different treatments and then follow them up and record the outcome
describe the crossover types of randomised controlled trials
- Patients are randomised into two groups, group A and Group b
- they are given the first treatment and then the outcome is recorded
- then they are swapped and given the different treatment and then the outcome is reordered
when is the parallel type of randomised controlled trials used
Parallel group – when effect of treatment is not reversible
When is the crossover type of the randomized controlled trials used
- Cross-over – when effect of treatment is reversible
What are the advantages and disadvantages of the cross over trials
Advantages
- Each patient is their own control
- Smaller sample size to get same number of observations
- Better for subjective measurements
Disadvantages
- More time consuming
- Carry-over effects – carry over effect of one treatment into other treatment period - therefore there should be a time period for example a month when the patient does not receive any treatment so it doesn’t carry over to the next treatment phase
What is cluster randomised trial
- Don’t randomise individual but randomise pre-existing groups to one or two treatments these could be villages schools, general practises
What is the positives of cluster randomised trials
- Avoid contamination (all participants in the trial are affected by the intervention even if only some receive it)
- Enhances compliance
Give an example of cluster randomised trial
- Cancer screening trials often done as this
- May need widespread publicity and women allocated to control might hear about the screening and demand it
What are factorial trials
- Assess 2 interventions using the same number of patients as 1 intervention
- Usually little or no interaction between the two interventions
- To assess 2 drugs using the same number of patients as 1 drug
describe the phases in developing and evaluating a new drug
• PRECLINICAL – Non-human study
– In vitro and in vivo animal experiments to obtain preliminary efficacy, toxicity and pharmacokinetic information
• PHASE 0 – First in-human trials (not always done)
– Small number of subjects given subtherapeutic dose of drug to determine pharmacodynamics and pharmacokinetics
• PHASE 1 – Screening for safety
– Testing of drug on (usually) healthy volunteers for dose ranging.
– Determine whether the drug is safe to check for efficacy
• PHASE 2 – Assess efficacy and safety
– To determine whether drug can have a therapeutic effect
– May be designed as case series or randomised controlled trial
• PHASE 3 – Assess efficacy and safety
– Randomised controlled trial on large number of patients to determine what the therapeutic effect is
• PHASE 4 – Post-marketing surveillance
– Safety surveillance (pharmacovigilance
How long can a clinical study last
Preclinical to phase 4 can take 12-18 years
Why do all studies need to be registered
• All trials must be registered prospectively
– Journals will not consider trials for publication unless they are registered with one of the primary registries in the WHO network or the United States’ clinicaltrials.gov
What are some of the registries in the WHO that clinical studies can register with
– International Standard Randomised Controlled Trial Number (ISRCTN)
– European Union Clinical Trials Register (EU-CTR)
– Australian New Zealand Clinical Trials Registry
What are the advantages of using registers for clinical trials
- Assist in the planning of new trials
- Avoid unnecessary duplication of research
- Avoid subjecting patients to trials seeking evidence that is already available
- Encourage collaboration between research groups
- Facilitate optimal use of research funds by funding agencies
- Facilitate patients’ access to information and improve recruitment
- Improve opportunities for methodological research
- Reduce discrepancies between published results and original trial protocol
- Help to detect publication bias in meta-analyses
how do you work out the death in treatment group and risk of death in the control group
Risk of death in the treatment group = number of deaths in the treatment group/Number of patients in treatment group
Risk of death in control group = number of deaths in the control group/number of patients in control group
What is the relative risk of death in the treatment group compared to the control group
Relative risk of death = risk of death in treatment group/Risk of death in control group
if the treatment as to effect what is the relative risk
relative risk is 1
Define intention to treat
Comparison of all subjects based on the treatment group assigned, regardless of whether they complied
Define on treatment
Comparison of subjects who actually took treatment
What is the difference between intention to treat and on treatment
• “Intention-to-treat”
Comparison of all subjects based on the treatment group assigned, regardless of whether they complied
• “On-treatment”
Comparison of subjects who actually took treatment
What is the primarily analysis, intention to treat or on treatment
Intention to treat
Why do you need to maximise compliance
• Need to maximise compliance to ensure trails results are meaningful
What does poor compliance cause
• Poor compliance on an intention-to-treat analysis reduces ability to detect treatment difference (if one exists)
How can you maximise compliance
– Selection of patients (not too ill)
– Double blind design
– Run in period where all get treatment (to identify those who can’t tolerate it)
How do you know the number of patients needed to treat (How many patients would need to be treated to prevent one patient getting the disease/ disorder)
Work out the absolute difference in risk
Absolute risk = risk of treatment group - risk of control group
What is the sample size
- Specify number of people to recruit prior to starting the trial
How do you work out the sample size
- If too few participants may not detect a real effect: study does not have enough statistical power
- Number needed calculated based on some prior information (eg expect relative risk of death of treatment A compared to B to be 2)
- Power is usually fixed at 85% or higher
if it is not statistically significant then we don’t need to determine if it is …
if it is not statistically significant then we don’t need to determine if it is clinically significant
How to determine if it is clinically significant
if the percentage difference is large enough
What is the purpose of a meta analysis
Purpose to bring together all the evidence to more powerfully estimate the effect size
What studies can meta analysis be used in
• Can be done for cross-sectional, case-control, cohort studies and randomised trials
What are the results of a meta analysis summarized in
• Results of individual studies and a summary estimate often shown in a Forest plot
What are the issues in meta analysis
Heterogeneity
Publication bias
What is hetrogeneity
Heterogeneity in statistics means that your populations, samples or results are different.
What causes Heterogeneity
• Caused by
– difference in study design
– difference in participant characteristics
– difference in intervention eg drug dose
– chance
• Can be assessed statistically – is there greater heterogeneity than you would expect by chance?
– Look for causes
What is publication bias
• Studies with significant or favourable results more likely to be published
What is publication bias caused by
• Caused by
– investigators
– journal editors
– journal peer reviewers
How can you assess publication bias
• Can be assessed statistically and graphically – funnel plot
How do you work out funnel plots
Funnel plot
- Relative risk of horizontal
- Standard error associated with relative risk of the vertical standard
- If there is no publication bias the points will be scattered either side of the line
