EBM and association - lecture Flashcards
Role of EBM
suggesting clinicians should use critically appraised information in clinical practice for optimal care of their patients
use current best evidence to make decisions about individual patients
Criticism of EBM
academic exercise for medical students - no relevance
not appropriate for GP
impossible to have the time
Governments, healthcare commissioners and providers used the jargon to justify decisions seen as inappropriate by clinicians
Hierarchy of evidence
systematic reviews and meta-analysis Randomised cohort trials cohort studies case-control studies ecological studies descriptive/cross sectional studies case report/series
Why is EBM important
patient care, safety, medical knowledge, professionalism
does not replace clinical decision making
clinical findings - how to gether and interpret info from history and physical examinations
aetiology - how to identify causes for disease
clinical manifestations of disease - know how often and when a disease manifests
diagnostic tests - how to select and interpret tests
prognosis - how to estimate it
therapy - sselect treatments that do more good than harm
prevention - how to reduce chance of disease and modify risk factors
Where can problems arise with randomised control trials
compare against placebo instead of current bets therapy
use inappropriate controls - might suggest more side effects or suggest it is less effective
if the trial isn’t published because it didn’t produce results eg Tamiflu - cant make decisions if can’t get all the info.
Example of evidence based medicine - cot death
case control study comparing baby’s last sleep and other factors like maternal medical history and medical history of the baby
10x more likely to get cot death if lying on front - odds ratio
back to sleep campaign
since 1991 cot death fallen by 75% - one of most successful public health campaigns
What is an association
link, relationship, correlation
statistical dependence between 2 variables
rate of disease of people with exposure is higher/lower than those without exposure
a result of chance, bias, confounding, cause
Describe chance
most studies based on an estimate from samples
assessed through statistical significance tests and confidence intervals
small sample size
describe confidence intervals
range within the true value is expected to lie within a given degree of certainty
samples are taken from population, confidence interval calculated for each sample then a certain percentage of the intervals will include the true underlying population parameter
describe the P-value
the probability that a result could be due to chance
0.05
P>0.05 result could be due to chance
Discuss bias
systematic error
incorrect estimate of effect of exposure
defects in study design or execution of the study
cannot be controlled in analysis/by sample size
2 types of bias
selection - systematic difference between those selected and those not, loss to follow up (attrition), non-response bias
measurement - measurements/classifications of disease are inaccurate, recall bias
describe confounding
any factor that has a real effect on risk of disease and related to risk factor under investigation
eg smoking and lung cancer - direct causal
proxy measures of more direct causes eg age and social class
common confounders
age
sex
socio-economic status
geography - N S divide
Describe causation
judgement based on a chain of logic that addresses 2 main areas
observed association is valid
totality of evidence taken from many sources - supports judgement
Bradford hill criteria
temporal relationship - exposure before outcome plausibility consistency with other investigations strength of association dose-response relationship - more exposure = more chance of illness specificity experimental evidence coherence analogy - difficul
Calculating odds ratio
lie on front case/lie on back case
lie on front control/lie on back control
case/control
What is relative risk
incidence in the exposed group/incidence in non-exposed group
exposed = non-circ
unexposed = circ
relative risk = 9.18
risk infection in non-circ more than 8x higher than in circ men
evidence to suggest causation - circumcision
overlying maps of circumcision and HIV - ecological study
experimentation - infection of ‘live’ human foreskin
case control and cohort observational studies
meta-analysis - calculate combined estimate of effect from multiple studies
attributable risk
rate of lung cancer among smokers-rate of lung caner against non-smokers
idea of how many extra cases exposure is responsible for
makes assumption exposure is causal
help guide policy makers in planning public health interventions
what is population attributable risk
a measure of the risk of outcome in the study population which is attributable to the exposure of interest
attributable risk definition
measure of exposure that indicates on an absolute scale how much greater the frequency of disease in exposed is compared with unexposed
assuming causal
difference between incidence rate in exposed and unexposed
risk attributable to exposure of interest
attributable risk = incidence exposed - incidence in unexposed
