EBM Flashcards
EBM is
The application of the best available, clinically relevant research to clinical care, involves integration of evidence with clinical expertise and patient values
EBM in practice means
Asking relevant questions to identify what is important from all available info
Looking for evidence that is peer reviewed, research studies with rigorous collecting methods and analysed data
Interpreting evidence by critically evaluating, considering biases and stats, whether relevant
What is a case report, when helpful
A retrospective report on a single pt
Used for special interest cases on things not commonly encountered
Helpful if have unusual case and see someone has done a case report on similar as can contact to improve your care
What is a case series
Collection of case reports that are similar and published together
Often same disease outcome or treatment
Focus on the specific cases not general trend
What is an ecological study
Retrospective
Uses population data that is already available to investigate overall trends and associations within large groups of people- ie city, factory, big GP
Generally looks at exposure and outcome
Pros and cons ecological study
Pros: data already available so rapid and cheap and large scale
Cons: doesn’t control for confounding variables relevant to exposure/outcome
Those with outcome may not have been those exposed
What is a cross sectional survey
For exposure and outcome prevalence at a single time point of a particular defined population
Retrospective
Uses a questionnaire that may be filled by a survey/interview/by the dr not pt
Cons of a cross sectional survey
Inaccurate recall
Wish to please investigator so unreliable
What is a case control study
Identifies risk factors, NOT effectiveness if treatment
Retrospective
Looks for possible risk factors in the history of a case group not present in the control group
Take history from both case and control groups, compare and conclude whether something likely a risk factor
Cons of a case control study
Doesn’t account for confounding variables, other differences may explain outcome
Can try to match controls to cases but difficult to include all confounders
what is a cohort investigation
follows a group or population (the cohort) over time (decades/years)
may compare groups or just follow 1
observational, no intervention
usually large scale
ie follow smokers and non smokers over time to see outcome
prospective
pros and cons cohort investigation
allows to study things would be unethical to manipulate ie exposing to dangerous things
confounding variables may explain outcome though statistics can help
people may leave the study as is long
randomised control trial
gold standard of clinical research
prospective
used for evaluating effectiveness of treatment
large numbers, random selection obviates confounders and bias, high validity
pts with similar characteristics randomly assigned to control or treatment group, follow up, compare results. control group gets placebo, used to show significance of treatment
can’t use to study risk factors as unethical to expose
cross over trial
type of randomised control trial comparing 2 or more treatments
each pt receives 2 or more treatments
tA, washout,tB
tB, washout, tA
pt randomly allocated to 1 or other treatment, follow by washout period, follow by next treatment
order of treatment may influence results
higher burden on pt so less compliance more likely to leave study
long lasting interventions unsuitable
systematic review
extensive critical analysis of all relevant publications
uses a framework and clear criteria for inclusion and exclusion
precise search term
methods section
takes a long time so may miss new stuff
meta analysis
technique for combing data from multiple studies
clear statistical analysis of pooled data
more powerful than its constituent investigations
looks at similar studies and analyses data for a statistically significant result
what are the retrospective study methods
case report/series
ecological study
cross sectional surgery
case control
what are the prospective study methods
cohort investigation
randomised control trial
cross over trial
what is the heirachy of evidence
- systematic reviews and meta-analysis
- randomised control trial
- cohort study
- case-control trial
- cross-sectional study
- ecological study
- case series and report
- ideas, editorials, opinions
what type of study is best for investigating risk factors
case control
what type of study is best to test a treatment’s effectiveness
randomised control trial
what type of study is best for a single time point of a defined population
cross sectional study
what type of study is best for following a group over time
cohort investigation study
what type of study is best for looking at exposure and outcome in large groups of people
ecological study
define validity
accuracy
validity is the extent something accurately measures what it should, the extent findings reflect the truth
define reliability
consistency
degree of stability shown when measurement repeated under identical conditions
define bias
when systematic error is introduced by selecting/encouraging one outcome above others
any deviation from the truth
define confounding variable
a third variable that influences both the dependent and independent variables
define independent variable
the variable that is manipulated by the researcher to cause a change in the dependent variable
define dependent variable
is affected by a change in the independent variable
give example of personal factors that influence how people perceive risk
emotional state
previous experience
education level
give example of contextual factor that influences how people perceive risk
how the risk is communicated
define relative risk, give example
relative likelihood of outcome in people exposed vs not exposed to risk factor
ie relative risk of thromboembolism if taking pill vs not taking pill is doubled, but reality is
if take pill= 2/7000
if don’t take pill= 1/7000
define absolute risk, given example
absolute risk is the risk of developing a condition over a certain time period
ie risk of thromboembolism if taking pill 2/7000, if not taking pill is 1/7000
change in relative risk is 1/7000
identify which of these is the absolute or relative risk
risk of eating a bacon sandwich daily increases risk of bowel cancer
there is an 18% increase in risk
it goes from 6% to 7%
18% increase in relative risk
absolute risk goes from 6 to 7%
which type of risk should not be used by itself
relative risk should not be used by itself
what are confidence intervals
margin of error, presented alongside risk info
often expressed as % where a population mean lies between upper and lower interval
