Interventional Studies Flashcards
What makes RCTs different from other studies?
Experimental intervention
Randomisation
Generalizability
In what ways are RCTs experimental?
One group is the control
The other receives an intervention
Types of randomisation
Simple
Block
Stratified
Minimisation
Quasi
Describe simple randomisation
Randomising each entrant separately e.g. using a computer generated list
What is block randomisation?
Randomisation occurs in a set of specified numbers e.g. blocks of 6 to ensure equal distribution between control and intervention group
Impact of larger block size in randomisation?
Lesser ability to predict allocation
What is stratified randomisation?
Baseline characteristics which may implicate final outcome are stratified so there is equal distribution between the groups.
Precision when using stratified randomisation?
Minimal if large study
What is cluster randomisation?
Unit of randomisation and analysis is various centers or catchment zones rather than individuals.
Significance of clusters in cluster randomisation
Within a cluster patients are more likely to respond in a similar manner and can therefore no longer be assumed to act as independent units.
Leads to loss of power
What is used in power calculations for cluster randomisations?
Effective sample size instead of actual sample size
What is used to calculate effective sample size?
Intracluster correlation
What does intracluster correlation represent?
Degree to which various individuals in a cluster resemble each other in outcome measure
Why is effective sample size needed in cluster randomisation?
People in a set cluster will act in a similar way to an intervention
When is minimisation randomisation useful?
When you want to do stratified randomisation due to worries of unequal distribution, but trial is too small to do so.
When conducting cluster randomisation and the clusters are small in number. You wish to minimise differences in confounder distribution
What can minimisation help do?
Achieve balance between treatment groups
What happens in minimisation?
Next allocation depends on characteristics of those already allocated
Advantages of minimisation
Ensures balance of prognostic factors between groups
Disadvantages of minimisation
Method is inferior to proper randomisation as allocation is exposed and controlled manually
Only first randomisation is truly random
Characteristics of sound randomisation
Reproducible order of assignment
Documented methods of assignment
Assignments remain concealed
Future assignments not predictable from past assignments
Ability to detect departures from established procedures
What is quasi randomisation?
Randomising using even/odd numbers of DOB/day of week patient was seen etc.
Problems of quasi randomisation
Not reproducible
Sequences cannot ensure equal distribution of variables
Advantages of randomisation
Permits use of probability theory in making inferences
Eliminates effects of bias
Facilitates blinding
Distributes baseline characters in unpredictable fashion
Improves generalisability
What does precision depend on?
Sample size
Not randomisation
Special types of RCTs
Crossover
N of 1
Factorial
Patient preference
Zelens modified
Non-inferiority
What is a crossover RCT?
Interchange of study and control groups after washout period so all subjects receive treatment and placebo
Which diseases can crossover RCTs not be used for?
Diseases cured by use of medication
What is parallel RCT?
Both control and intervention happen in parallel, not in serial fashion i.e. opposite of crossover
What is N of 1 trial?
Single subject is administered placebo and intervention in tandem under double-blind controlled conditions.
What is factorial trial?
In 2x2 factorial design, participants are allocated to one of four combinations (2 treatments, 2 levels).
How many possibilities in a 2x2x2 trial?
3 treatments at 2 levels
Advantage of factorial trial
Can test both independent effect of each drug and combined effect
What are patient preference trials?
Patients may have strong preference for certain therapy, and are allocated to that one.
Only those with no preference are randomised.
How can researcher obtain data from patients who have preferences in preference trials?
In parallel fashion
Advantages of patient preference trials
Increased recruitment
One can analyse motivation on intervention effectiveness
Disadvantages of patient preference trials
Recruiting for no preference can be more difficult
Obtaining consent canbe complex
What is Zelens modified RCT?
Randomisation occurs before consent; consent only from those randomised to experimental treatment.
Reason behind Zelens modified RCT?
In standard RCT, patients tend to withdraw if they suspect they are in the placebo group and not receiving treatment they hoped for.
Disadvantages of Zelens modified RCT
Ethically controversial; MRC does not accept this design as ethically valid
What are non-inferiority RCTs?
When aim is to show that new intervention is ‘‘not inferior’ to standard intervention
What is null hypothesis of non inferiority RCTs?
There is significant difference between two interventions and not otherwise as seen in superiority trials
Another name for non inferiority trials?
Equivalence trial
Disadvantages of non inferiority trials
Higher sample size needed compared to superiority trial for same level of power and significance
Intention to treat analysis may blur differences between groups and may support claim of equivalence
In which trials is it best to report both ITT and per-protocol analysis?
Non-inferiority
Types of interventional trials
RCTs
Uncontrolled
Before and after
Multi arm
What happens in uncontrolled trials?
Only one group observed for effects of intervention - no control.
What is study inference made from in uncontrolled trials?
Using historical control i.e. status of treated group before intervention
What happens in before and after trials?
Outcome is assessed before and after the intervention in one group of subjects.
Weakness of before and after trial?
No randomisation
Observed change may be due to a factor other than intervention
What is a multi arm trial?
RCT where there is more than one study arm.
Disadvantages of multi arm RCT
Larger sample size so power may be reduced
What happens to power as sample size increases?
Power reduces
What is efficacy?
How well a drug works under optimal circumstances
What is effectiveness?
How well drug works under usual practice circumstances
What is efficiency?
Measures whether healthcare resources are being used to maximise value for money
When is efficacy trial undertaken?
To meet regulatory approval
When is effectiveness trial undertaken?
Convince formularies and payers of actual usefulness of drug in current practice
What are pragmatic trials?
Effectiveness trials for which hypothesis and study design are designed to answer the questions faced by decision makers
Unnatural circumstances in RCT compared to regular clinical practice
Recruitment from specialist centres
Exclusion of patients with co-morbidities
Careful selection of patients to generate homogenous diagnostic groups
Random allocation
Detailed information given for patient consent
Placebo used to compare active treatment
Patients followed up frequent intervals
Patients receive detailed checklist of SEs
Assessment endpoint is usually 4-6 weeks
Sx measures are outcomes considered
Patients + clinicians blinded
Characteristics of pragmatic RCTs
Select clinically relevant alternative interventions to compare
Include diverse population of participants
Recruit participants from heterogenous practice settings
Collect data on broad range of health outcomes
What do pragmatic RCTs tell you (and not tell you)?
Inform how effective intervention is in real world
Do not tell you how efficacious t is
Problems with pragmatic RCTs
Substantial drop-out rates
Blinding can fail
Less rigorous methodology
Higher rates of non-specific treatment
May not employ placebos
May not be blinded
Must carefully conceal allocation during effects
Advantages of pragmatic RCTs
Reflect heterogeneity of patients in clinical practice
Minimise exclusion criteria
Focus on groups with wide range of diagnoses
Define patient groups by presentation rather than diagnosis
randomisation
May evaluate real world, function based outcomes
