70 - Clinical Trials in Respiratory Disease Flashcards
Top level of NHMRC evidence
Systematic review of randomised controlled trials
Bottom level of NHMRC evidence
Case studies
Key marker of airway obstruction
FEV1
PICOT
Population Intervention Comparator/Control Outcome Timing
Relevance of PICOT
Ideas for designing a clinical trial.
Fluticasone
Anti-inflammatory agent
Two broad areas for judging relevance of evidence
Internal validity and external validity
Internal validity
The extent to which the results of this study valid for the sample of patients being studied.
Questions for determining internal validity
1)
2)
1) How well did the study answer the question it set out to answer?
2) Dependent on appropriate study design, control for bias and confounding variables
Aims of randomisation
1)
2)
3)
1) Treatment groups all identical in all aspects other than the intervention
2) Even distribution of potential confounders (even those that are unknown)
3) The primary rationale is to reduce confounding
Stratified randomisation
Randomisation stratified by levels of key confounders (EG: randomise subjects within stratum to which they belong, EG: country and smoking status)
Make composition of groups even more similar with respect to key confounders.
Design to deal with key confounders
Stratified randomisation
Blinding (masking)
Non-awareness of intervention allocation
Rationale is to reduce information bias
Levels of blinding
Single blind - Subjects unaware
Double blind - Subjects, investigator unaware
Triple blind - Subjects, investigators, outcome assessors unaware
How should outcomes be assessed?
Outcomes determined by strict, standardised, objective criteria.
Centralise the assessment in multi-centre studies to improve standardisation.
Rationale is to reduce information bias
How to reduce intention-to-treat bias
Assume that subjects remained in allocated group, regardless of actual treatment.
Subjects who drop out, cross over, etc are almost always systematically different from those who don’t
Randomised control trial of drug versus placebo potential source of bias
sick subjects cease new drug due to side effects
• selects for healthier group in drug group, which
experiences less outcomes
• misperception: new drug better than placebo
Side-effect of intention-to-treat analysis
Always underestimates the treatment effect.
Less treatment in intervention group than assumed, more treatment in control group than assumed
Thing to look for in a clinical trial design
Intention to treat analysis
Sample
A subset of the whole population, from which assumptions about the whole population are made
P-value
Probability that the observed result is the result of chance.
Conventional P-value cutoff for statistical significance
Under 0.05
Null value
Value if there were no difference between the groups being compared
EG: 1.0 for ratios (HR, RR), 0 for absolute differences.
Number needed to treat
Number of people needed to undergo intervention in order to prevent an outcome in one.
NNT = 1/(absolute risk or rate reduction)
A marker of the efficacy of the intervention
Number needed to treat (the lower the better)
Things affecting number needed to treat
Relative effect (often constant) Underlying likelihood of outcome (the lower, the lesser the NNT)
What does the effect of underlying likelihood of outcome on NNT say about efficient treatment?
Relative benefits don’t tell everything. If a drug reduces an outcome by 50%, but the outcome appears in 0.5% of the population, NNT will be really high. This shows that the intervention mightn’t be so effective.
External validity
How well the results of the trial be generalised to the wider population
How can external validity be assessed?
See how concordant the randomised controlled trial subjects and clinical subjects are (according to PICOT)
