before evidence based medicine…
- many medical treatments were either ineffective or harmful
- e.g. blood letting
- however, some older medicines based on evidence
standardised evidence based medicine (EBM)
recent movement (1980s and 90s) which emphasizes the importance of unified standards and ranking of evidence across medical practise research
what does EBM target?
over reliance on:
- clinical experience and judgment
- background theory
EBM recommendations
- update knowledge of medics regularly
- make results of clinical trials more widely known
- fix what constitutes best available evidence (aka evidential weight)
randomised controlled trials
- experiment where half of the people are given treat, half given placebo
- without RCT no real medical evidence
examples of therapies which do not have RCT evidence but still work
- penicillin for bacterial pneumonia
- diuretics for heart failure
- appendectomy for appendicitis
more sophisticated reading of EBM?
- other evidence type may be available and should be consulted
- however, RCT trumps them
why are some sources of evidence ranked higher than others?
imposing further contraints on source
evidence hierarchy from best to worst
- meta-analyses and systematic reviews
- RCTs
- cohort studies
- case reports
- expert opinion
how do we make a combined judgment when evidence comes from different sources?
- by assigning weights to evidence
- issue is how to determine weights in non-arbitrary way
why choose RCTs?
- in any study, want to ensure that any effect attributed to specific action
- problem = cannot know what would happen to same objects/subjects in absense of action
- overcome problem use a comparison group which isn’t subjected to action
- treatment group
- control group
- group subjected to specific action
- group not subjected to specific action
subjects in both treatment and control group must:
- possess the same conditions under study (e.g. same illness)
- be differently subjected to actions (e.g. treatment vs non-treatment)
issues with RCTs
- individuals in control group may be suffering from more aggressive form
- individuals in control group may be frailer
how to mitigate RCT issues?
- remove subjects with confounding factors from study
- equally distribute factors by intential selection
what is the major obstacle to make everything equal?
- may be unknown confounding variables (internal validity issues)
- randomisation meant to solve this because unbiased method of distributing subjects into groups (hopefully even distribution of confounding variables but no guarantee)
internal validity
when treatment’s impact accurately measured in study population
