measures of association Flashcards
what types of studies are medical research studies?
they are RCTS, case-control and cohort
how are outcomes measured?
a variety of ways - categorical (positive or negative), continuous subjective measures such as pain, and continuous objective measures such as HbA1C
what are measures of association for?
the effect of new treatments with dichotomous outcomes
how can the measures of association be measured?
relative risk, risk difference or odds ratio - they will show the direction and magnitude of association between variables
what is probability?
it is the measure of chance of getting the outcome of interest from an event
what is the equation for probability?
it is the number of times the outcome of interest occurs / the total number of outcomes possible
how do you calculate the percentage chance?
it is the probability x 100
what is risk?
it is a way of expressing treatment effects - it is the probability of an event occurring
how do you calculate absolute risk?
the number of events / the number of people at risk
what is relative risk?
it is the ratio of the risk of an event in a particular group e.g. exposed in one group compared to the risk of outcome in an unexposed group (in RCTs - treatment group compared to the control group)
what traditionally is the numerator?
the group of interest. The denominator is the unexposed or control group
explain RR = 1?
it means that the risk of outcome is the same in the exposed as the unexposed group
explain RR <1 and RR>1?
RR<1 means that the risk of outcome in the exposed group is less than the risk in unexposed and >1 means that the risk of outcome in the exposed group is higher than that in the unexposed group
what is the range for ratio values?
0 to infinity
how would you obtain percentage reduction?
subtract the RR from 1 and x100
what does a large value of RR imply?
it implies a strong association - must always be interpreted in association with the confidence interval
what is the odds?
it is the probability of the event occurring / the probability of the event not occurring - number of times event happens / number of times it does not
what is the odds ratio?
it is the odds of the event in one group / the odds of the event in another group (exposed/unexposed)
what does OR = 1 mean?
the odds in the exposed group is the same as the odds in the unexposed group
what is 95% CI used for?
it is used to estimate the precision of the OR and the RR
what does a large CI mean?
low precision of RR or OR
what does the CI crossing 1 mean?
there is not statistical significant difference between the two groups - exposed and unexposed
why should RR not be used in retrospective or case control studies and what are they used in?
they are not used in these because they are not meaningful - the proportion of cases or control is chosen by the researcher. They are used in RCTs, cohort and prospective studies
what would you use in case-control or retrospective?
OR - commonly reported in logistic regression analysis
what happens if the disease is rare?
RR and OR are similar
what is the risk difference?
it is the difference in risk of outcome in the exposed and unexposed or treatment and control group
what would a difference of 0.48 mean?
there is a 48% increased/decreased risk for developing outcome for one of the risk factors or variables
what is useful about RD?
it is meaningful to find out if findings from a study are worthy of being incorporated into clinical practice - magnitude of risk difference may or may not warrant changing practice
what is NNT and what is it used for?
it is number needed to treat and it is a way of evaluating the clinical importance of treatment in practice
how do you calculate NNT?
1/RD
when is NNT used?
in studies comparing an intervention and standard treatment
how would we describe NNT simply?
it is the number of people that need too receive a treatment in order to see the desired effect in at least one of them
how would CI contraindications relate to medical practice?
if there was a contradictory measurement in CI such as crossing 1 then would be concerned. Likewise be concerned with your diagnosis if the consultant would have diagnosed with something different 95% of time
what are CI?
they are values to show that 95% of the time a mean will lie in this range
what is the p value?
it is the probability of a coefficient being at least as big as yours - assuming the coefficient is 0
what does a small p value represent?
the zero-assumption is probably wrong and then effect is likely and vice versa for a large p value - evidence is not strong enough to suggest an effect
how would you carry out statistical process?
observe, guess, test, assess
what is R squared?
how good the fit of the model is - how well the points align to the line of best fit
what is S?
it is the deviation of the points