AAMC 5 Flashcards
what is critical appraisal of research
apply rules of science and common sense in judging the quality of written articles
6 standards for evaluating quality (FiLCHeRS)
Falsifiability - can the results be proven wrong (no god)
logic - are they logically coherent
comprehensiveness - evidence offered must be exhuastive, all evidence is presented and don’t ignore contrary evidence
Honesty - claim must be evaluated with an open mind and without self deception
Replicability - time and populations
Sufficiency - evidence offered must be adequate to establish truth to that claim - extrodinary claims require extrdinary evidence
what is EBM
integrate clinical experience with the best available external evidence from the literature
5 As of EBM
Assess: recognize and priotize problems
Ask: construct clinical question that facilitate efficicent search
Aquire: gather evidence from quality soucres
Appraise: evaluate evidence
Apply: to inidivuals, taking into account their prefs/values
what is diff. in hypo between qual and quant
qual - hypo generatine
quant - hypo testing
what diff between inductive and deductive
indctive (qual) - works from instance to generalize
deductive (quant) - works from the general to the particular
3 quesions to evaluate qual research
- was the design rigorous
- was study execution rigorous
- can I tranfer results to my setting
7 factors to infer causal relats
- chronology
- strength of assoc.
- dose response
- specificity of assoc. - agent or risk only relates to this disease
- consistency - acrsoss studies and pops
- coherent or plausible - is there a reasonable explanation
- cessation of exposure - if remove does it go? best for population
5 steps of an experimental study
- formal hypothesis stated
- select people for exposure
- sample divided into 2 groups -can be:
a. randomized (RCT)
b. non-randomized (quasi experimental - inferior) - one group given intervention while other group not
- outcomes recorded and compared
2 ways to administer intervention in RCT
single blind- participant doesn’t know
double blind - patient and researcher don’t know
what is triple blind
stats person doesn’t know
limits of RCT
- very controlled conditions
- often very select populations
diff. between efficacy and effectivness
efficacy - impact in optimal conditions
effectiveness - impact in real world
3 ethical considerations to RCT
- unethical to deny treatment, so can’t know if one is better
- unethical to adopt new treatment without testing before
- unethical to continue trial if treatment is found to be good
4 phases of intervention study
before - animal studies
1 - tested in small group of healthy to assess safe dose and safety/SE
2 - given to a larger group at reccomended dose to determine efficacy and safety
3 - drug given to a larger group to confirm effectiveness, monitor SE and and compare to other treatments (often mutliple RCTs)
4. post-market surveillance
what is major problem with obs. studies
hard to infer causation
2 types of obs. studies
- descriptive - simply describe how thing are - usually cros sectional
- analytical
def. analytical study
- designed to test hypothesis predicting an association
- 3 types
- cross-sectional
- cohort
- case control
def. cross sectional study
- subjects selected irrespecitve of presence or abscence of chars. of ineterst
- similar to descriptive, but tests assoc. hypothesis
def. cohort study
group of people that can be sampled and enumerated over time that share a defining characterisitc
- usually look at people with an exposure and statrt before they have a disease
how is cohort study analyzed
gives you grid and can use RR to assess (incidence in exposed/incidence in not exposed)
main adv. of cohort
exposure recorded before ourcome, so can establish temporal aspect of causality
def. case control study
compare group with a particular outcome to a similar group without the outcome
how is case control analysed
with OR (ad/bc)
what is problem with giving RR
can seem scary and increase of 100% is not a big deal in rare disease (1/100000 becomes 2/1000000)
what is attributable risk
takes into account the magnitiude of risk - indicates the number of cases of a disease among exposed indivs that can be attributed to that risk
formula for attributable risk
incidence in exposed - incidence in unexposed
what is exposed attributable fraction
proportion of the incidence in the exposed person caused by this factor
- eg. 89% of your lung CA risk is due to smoking
formula for EAF
(incidence exposed-incidence unexposed)/incidence exposed
formula for NNT
1/ARR
2 general categories of bias
- sampling (and non-response)
2. measurment
3 types of measurement bias
- social desirability
- instrument problems
- recall bias
what is information bias
objectivity of the investigator
what is confounding
thrid variable fucking shit up
what is heirarchy of evidence
1 - at least one proper RCT
2a - well designed controlled trials without randomization
2b - well designed cohort or case control trials, prefer more than one center
2c - multiple time series studies, with or without intervention
3 - expert opinion, descriptive studies, consensus conferences
quesitons to ask self about research
- are results valid? - study design and analysis
- what are results? size and precision
- can I apply to PT care? - similar PTs and env.
- what do I do with this PT?
4 things to consider
- target population
- intervention - feasable? time?
- cost
- placebo effect
