Understanding clinical trial data Flashcards
how do you demonstrate evidence of whether the product works
- whether they have taken a lead activity molecule
- made structural modifications to achieve best possible PK and PD properties
- formulated it into an elegant and acceptable medicine
how do you demonstrate efficacy
- have to demonstrate that new product produces desired effect
- provides the means for a therapeutic intervention that will benefit a patient
- including tolerability and acceptability - evidence has to be capable of being verified
what does an ARR of 0.08 mean
indicates the reduction of risk by 8%
- eg. patients is 8% less likely to have a stroke
what is meant by numbers needed to treat
the number of patients who would have to receive treatment for 1 of them to benefit
what evidence does clinical trials require
evidence requires a comparison between a test group and a control group
- control group can receive a placebo or another form of treatment
what is a placebo
an inert substance or dosage form which is identical in appearance, flavour and odour to the active substance or dosage form
- used as a negative control in a bioassay or clinical study
when would the use of a placebo not be used
ethics
- not treating certain conditions would be unethical
- hypertension, cancer, pain
what does statistical analysis involve
- requires a comparison between 2 groups
- uses calculations to establish whether the behaviour of 2 groups is different
what is a null hypothesis
the claim that there is no effect in the population
what is the statistical conclusion
the outcome of the trial should disprove the null hypothesis
how is a clinical trial set up
- recruit a number of patients with a condition
- divide them into 2 groups- test group and control
- compare outcomes at the end of the trial
how can trials be flawed
- statistical errors
- hypothesis
- statistical set up
- analysis of data - bias
- setting up the groups (selection bias)
- observational bias
- confounding factors
what is the central assumption made on results
the outcomes from a trial will be applicable to the treatment population as a whole
how is the trial population identified
- must mirror the treatment population
- need to have criteria in place to match the trial population to treatment population
- nature of condition
- inclusion and exclusion criteria - demographics
- age, gender, ethnicity
how is the inclusion criteria identified
- identify the condition to be treated precisely
- use clinical diagnosis and standardised measurement scales
how is the exclusion criteria identified
- exclude those for who the treatment wouldn’t apply
- identify factors that might interfere with results
- comorbidities
- other medications
- factors affecting adherence to protocol
what is a normal distribution
- a population, when measured against a criterion, will usually show some level of variance
- this variance is likely to show a distribution around an average value
how is the normal distribution identified in a clinical trial
- trial population must mirror treatment population in all criteria
- trial population is divided in 2 groups
how is randomisation ensured in a clinical trial
- for the outcome to be true, the test group must be the same as the control group in all respects
- can use random number generator
- should be done by an independent person not involved in the trial
what is meant by blind and double blind
- the operator shouldn’t know what treatment an individual has had
- highlights importance of placebo - double blind- neither the operators or the patients know the treatment
what is a cross over trial
- half way through the trial, the test and control groups are swapped over
- requires measurements all the way through
- trial usually runs for longer period of time
- trials can be stopped if it becomes clear that one treatment is significantly better than the other - applicable to certain conditions- pain relief, insomnia
what is the power calculation
- trial must be designed to be capable of disproving the null hypothesis
- needs to be a large enough size to show that outcomes for the 2 comparison groups are statistically different from each other
- large enough to eliminate play of chance
what is the power of a trial determined by
- what kind of statistical test is being performed
- sample size- larger the size, larger the power
- size of experimental effects
- level of error in experimental measurements
what is intention to treat
- includes all randomised patients in the groups to which they were assigned
- a high drop out rate compromises the power of trial and could hide confounding factor
- ITT reduces risk of bias in analysis of data
what is on treatment analysis
- seen in older clinical trials
- analyse data from patients who complete the trial
- drop outs not accounted for
how is consistency achieved in data collection
all participants must be treated in the same way
- data collected according to same protocol and at same times
- deviation can lead to bias
how is precision achieved
data analysis should give statistically meaningful indications of the precision obtained
- confidence intervals
- coefficient of variance
describe the impact of the outcome of a trial
- outcome should relate to the overall health benefit of the therapy
- RRR
- ARR
-NNT
how can adverse effects be measured
numbers needed to harm- number of patients who would have to receive the treatment for 1 of them to experience an adverse event
