Stats Flashcards
What is a p-value?
The p-value is the probability (0-1) that the outcome of an experiment was due to pure chance, as opposed to an actual difference.
P-value = the probability that the null hypothesis is true
Most papers aim for a p-value of <5% or <1%
What is power?
The likelihood of a study detecting a true difference.
This involves a calculation that determines what sample size will be needed to say a detected outcome is statistically significant.
What is a confidence interval?
The confidence interval is the range in which you are 95% sure that the real population result lies somewhere between
What is absolute risk?
The probability of an event (or non-event) occurring within a specific group.
An individuals risk of developing a disease over a given time period.
What is relative risk?
Risk of event in intervention/ exposure group
Divided by risk of event in control group
Ultimately, relative risk compares the risk across different groups
What is number needed to treat (NNT)?
The number of people that need to take a treatment for one person to benefit from the treatment.
Can be calculated quickly using
NNT = 100/ %reduction in absolute risk
What is an odds ratio?
The odds of an outcome occurring if exposed to something, compared to the odds of that outcome occurring if unexposed to that same thing.
Odds ratio is often used in case-control studies.
What is a Hazard ratio?
At a specific point in time, the probability of an event (often a negative outcome) occurring in the intervention group compared with (divided by) the same probability of that event occurring in the control group.
What does a Kaplan-Meier curve show?
This is a graph that shows a ‘time to event’ analysis, with the event often being death or an alternative (surrogate) clinical endpoint.
The plot appears to follow a trend of down going steps, with each step corresponding to an ‘event’ occurring.
Censored events = time at which patient withdrew from study, or trial finished.
What is a single blind and double blind study?
If participants of a study are blinded to their treatment allocation (intervention vs control), this is known as a single blind study.
If the experimenters allocating treatment, measuring data and analysing it, are blind to the group allocation, this is said to make the study ‘double blind’.
What are confounding variables/factors?
A variable/factor that affects the dependent variable but is unaccounted for. This can be reduced by randomisation, blinding and matching groups as best as possible to align baseline characteristics.
What is intention-to-treat analysis ITT)?
Analysis of trial participants as per the group they were originally allocated to (irrespective of whether they were non-compliant, or withdrew from the study etc.)
What is per protocol analysis (PP)?
Analysis of trial participants with exclusion of those who did not follow protocol (non-compliance/ withdrew from study etc.) This is to identify the treatment effect under optimal conditions.
What is internal validity?
The trustworthiness of a study which is seeking to establish a ‘cause and effect’ relationship, earned through robust methodology and elimination of systematic error/bias.
What is external validity?
This is the validity (or generalisability) of the study in the context of the wider population.
A study’s exclusion criteria for patients may be a good start for evaluating external validity!
