Evidence Based Dentistry Flashcards
what is risk
chances of something happening - number of times something happened over the total number of times it could have
what is risk ratio
between two groups - control and clinical. the risk of clinical over the control
what is the absolute risk reduction
the difference in risks between two groups - control and clinical
if the risk ratio is 1, what does this suggest
that there is no difference between the risks of each group, therefore there is no difference in outcome between the control group and clinical group
what is the absolute risk difference
the difference between the risks of the two groups, the risk reduction. if this is 0 - no difference in outcome of the groups
what is a confidence interval
applying the risk ratio to the full population, instead of a sample population, the range of values of risk ratio is likely to lie when in a true population
when can confidence intervals suggest insufficient evidence for an outcome
if the confidence interval overlaps 1, the risk ratio could be 1 meaning the risks of outcome between the two groups could be equal, therefore there is no difference
when can a confidence interval suggest sufficient evidence for an outcome
if the CI does not overlap 1, either greater than or less than 1, there is a difference between the risks of outcomes between groups
if difference is looked at, what suggests insufficient evidence
a value overlapping 0 - no difference in values
describe a case study
when one patient or a group of patients are investigated, noting symptoms, treatment and outcome but not controlling anything, just reporting
describe a cross sectional study and what is it used for
when patients are looked at at a point in time, noting the number of patients with disease and on treatment, can be useful for prevalence of disease or identifying risk factors, only observational
describe a case-control study
having two separate groups - one with disease and one without. then investigating any differences, looking at risk factors. useful for looking at potential causes of disease, but can have recall bias
describe a cohort study
identifying a population - e.g. pregnant mothers - then following these participants, recording behaviour, measure exposure eg smoking, then looking at outcomes or diseases, e.g. premature birth, or congenital defects. very time consuming, might need a large sample size
describe a randomised control trial
having two separate groups and controlling all other variables. one clinical group and one control, randomly putting participants in each group and measuring outcomes. best way for causal effect, strong evidence on effectiveness of treatment
describe the four design elements of RCT
participant criteria - need to have an equal base line to start with, age, gender, disease
randomisation - participants should be randomly put into a group, done by a computer to avoid any bias.
control/comparison groups - placebo or current standard
blinding/masking - participant and investigator should not know what group they are in, knowing this might skewr results with bias
