Randomisation, Blinding And Protocol Deviations Flashcards
Define allocation concealment
Not revealing to the participants what group they have been allocated to
CONSORT
Consolidated, standards of reporting trials
CONSORT guidelines flow diagram ensures?
All potential ppts are accounted for
Everyone recruited is accounted for
Numbers a located to each ‘arm’
Number of drop out (and why)
Number of computers
CONSORT allows?
US to see any obvious differences in number of participants between arms (eg groups) especially group allocation and drops outs
Steps in CONSORT
Enrolment
Allocation
Follow up
Analysis
Criteria for selection
Need people who have disease
Select population by choosing people who don’t have certain conditions
Often exclude: Old, children, pregnant and who don’t consent
Advantages by simple randomisation
Can’t predict in advance who gets control and treatment
Control group roughly resembles treatment group
Not just characteristics you know about but also characteristics you don’t know about
Disadvantages by simple randomisation
Can’t guarantee same number of participants in each group, especially if small study
Some factors that predict outcome of study may not be evenly distributed between the groups
Advantages simple randomisation / block /stratified
Simple- reduces selection bias
Block - reduces selection bias and prevents unequal treatment group sizes
Stratified - reduces selection bias and prevents imbalance in prognostic characteristics
Disadvantages simple randomisation / block /stratified
Simple- groups can be unequal unless trial very large
Block - imbalance in prognostic characteristics car still occur
Stratified - more elaborate and need to be certain about what characteristics are predictive
Different types of blinding
Allocation concealment
Single blinding
Double blinding
Triple blinding
Allocation concealment
Allocation of each participant not revealed until irrevocably entered trial (minimises selection bias)
Single blinding
Usually participant kept unaware of assignment
Double blinding
Keeping both participant and those involved in their management unaware of assignment
Triple blinding
Keeping participant and those in involved in their management and performing blind data analysis
How to do binding
Conceal allocation
Mask name of treatment
Think about appearance, taste, smell, colour and mode of delivery
If you have intervention + standard treatment, then may want to consider placebo + standard treatment
Advantage of blinding for assessors
Minimises differential assessment of outcomes ( ascertainment/ information/ observer/ detection bias)
Advantage of blinding: participants
Less likely to have biased psychological /physical response to intervention, especially it subjective measure e.g pair score
More likely to comply with treatment
Less likely to seek additional interventions
Less likely to leave trial early
Advantage of blinding: investigators
Less likely to share inclinations
Less likely to seek additional interventions
Less likely to leave trial early
Disadvantages of blinding
Less important for objective criteria (e.g death)
Classes for protocol deviations
Major
Moderate
Minor
Major protocol deviations
Failure to obtain informed consent
Failure to report serious adverse events
Failure to perform procedure that affects safety
Dosing error
Moderate protocol deviations
Use of outdated/expired/unapproved consent form
Enrolment of ineligible participant (e.g outside age range)
Study visit outside of required timeframe
Minor protocol deviations
Loss of original signed consent form or inappropriate documentation of informed consent
Failure to give participant copy
Over-enrolment
Pk sample obtained close to, but not as, assigned time
Intention to treat analysis
All study participants taken into account in analysis, whether completed study or not
Intention to treat analysis is analysed according to ?
Group to which they were allocated
Intention to treat analysis keeps advantages of?
Randomisation (known and unknown features remain evenly distributed) and results more generalisable
Intention to treat analysis: conservative approach
Underestimates benefits of interventions, avoids overestimation of treatment effects
Per protocol analysis
Only takes account of participants who always adhered to study protocol I.e. always received intervention and completed follow up
All participants with protocol deviations excluded
Per protocol analysis disadvantage
Can overestimate treatment effects ( people who stop treatment due to no benefit not included)
Other types of per protocol analysis
As-treated analysis
Completers-only analysis
As treated-analysis
Participants evaluated as ultimately treated regardless of assignment
Based on actual treatment received (some people may crossover between treatment groups)
What per protocol analysis is useful to analyse side effects
As treated analysis
Competers-only analysis
Only those who complete study included
Which Per protocol analysis is used to manage participants with missing data?
Computers- only analysis
