EBM Day 5 Flashcards
when not best to perform RCT (4)
- unnecesary
- inappropraite
- impossible
- inadequate (not best test design)
equipose
not sure which treatment is better
RCT sampling
decrease variation
- reduce internal error
- incrae internal validity
- reduces external validiy
exclusion criteria
comorbidty limited longevity adherence/retention concerns non-cooperation highly selected-not much like everyday patient
random allocation
avoid potential confounding
factors related to prog are distriburted if groups are large enough
problems after randomization
cross over
co interntion
nonadherence
4 levels of blinding
allocation concealment
blinding subjects
bliding treating physicians
blining investigators
Open trial
no blinding
ideal for test that can’t have blinding (surgury)
Outcomes
primary
secondary-too many cause t1 error
composite-evaluate secondary components sep and together
NNT
1-abs risk reduction
How many patients one would we need to treat, on average, to prevent one outcome event
Abs risk reduction
incidence affected %- incidence unaffected%
Efficacy vs effectiveness
does trx work under ideal conditions vs under circumstances of care
Intention to treat vs per protocol analysis
Analyze in group should be in
- preserves original randomization
- answers which trx choice is best
analyze only who completed study per protocol
- answers does actually taken trx instead of being offered make better
- concern about bias
superiority, equivalnce, noninferiority trials
equivalence is different in any direction
Type 2 error causes
small sample size, poor adherence, co-interventions, crossing over, loss to follow up , differential/nondifferential measurement error
Cluster randomized trials
Naturally occurring groups of patients are randomized as a unit
Cross over trails
each patients get all trials after wash out period
controls for difference in response to trx among patients
Practical for outcomes that recur and trx is transient
Prevention levels
primary-before disease
secondary-after exposure
teritary-after disease is recognizable
quay-too much trx-hurting now
Commune
clusters of unvaccinated susceptible children or adults
Vaccine controversey
morality, necessity, safety
autism-time around when autism presents
also remve the mercury from vaccine in US-must keep refrigerated
Herd protection
must have 90% vaccination rate to work
1 with disease-disease mutate-vaccine no work
Lead time bias
illusion screening-detect test earlier indicating prolonged survival
Pricipals for screening disease
High prevalence problem that is understood accepted trx with means to trx suitable latent and symptomatic stage suit test for diagnosis acceptable agree whom to treat cost of finding economically balanced with overall health continuous process-
Lead time bias
discover disease early on-patient lives longer
makes screening look effective
length time bias
slower killing disease keep you alive longer
makes screening look effective
Compliance bias
people who participate in screening likely to participate in follow up
makes screening look effective
Slow growing cancers and screening lead two which error
false positive-type 1 error
screening in low prevalaence diseases
not done
few true positives
many false positives
PPV is low
True negatives are high
-follow up on positive screen (ELISA and WESTERN)
-if second test is biopsy many complications can occur
chemorprevention
usage of drugs, diet sups to reduce disease chance
USPTF
US preventative task force
read literature and say what screens are worth it
says evidence is insufficient a lot
