Evidence Based Medicine Flashcards
systematic deviation from the truth (not due to chance)
bias
sample population has an equal chance of being assigned to the intervention or control group
spreads confounding variable makes the prognosis more similar across groups
increases the likelihood that the intervention is the only difference between groups
randomization
investigator cannot influence the allocation at the time of study entry
allocation concealment
participants are analyzed in the groups to which they were randomized even if they did not get the intervention
cross-over introduces bias
intention to treat
poor accuracy, definite causes, reproducible
selection bias, information bias, confounding
systemic error
poor precision, nonspecific causes, not reproducible
random error
the number of people experiencing an event as a proportion of the total number of people in a population
absolute risk
arithmetic difference between two event rates
risk of the adverse outcome in the control group = baseline risk
control rate - experimental rate
absolute risk reduction
proportion of the baseline risk that is still present when patients receive the experimental treatment
experimental rate/control rate
relative risk
what proportion of the baseline rate has been reduced by the experimental treatment?
1 – RR = (control rate – experimental rate)/control rate
Relative risk may over-represent the effect of the intervention if the baseline risk was low
relative risk reduction
the number of people that would have to be treated to help one person
100/ARR
number needed to treat