Bias- Block 1 Flashcards
what are the 4 major types of information biases in cohort studies
recall
observer
detection
latency and accumulation
what are the 4 major types of selection biases in cohort studies
volunteer/non-response
referral
drop out
healthy worker effect
what is recall bias
experimental subject gives false information
what is observer bias
incorrect records
what is detection bias
one group is examined more closely than the other group for outcome
what is latency and accumulation effects
failure to account for induction period of a disease and the fact that dose accumulation can have an increased effect
what is volunteer/self selection bias
people are more likely to participate in a study they have more concern about
what is referral bias
ex: people have to be referred by primary care to a specialist who recruits them into a study
what is healthy worker bias
people with jobs tend to be healthier than those without (exposed group may be more healthy than the unexposed group)
what are the 3 ways in which we can control for confounders in the design
randomization
restriction
matching
how can we control for confounders in analysis
stratified analysis
why are prospective cohort studies more reliable than retrospective
you eliminate the chance of recall bias
what bias is most significant in case control studies
recall bias
in case control bias, are incident or prevalent cases better to use
incident
what type of bias may result from using prevalence instead of incidence cases for a case control study
late look bias (the most severe cases may have died
or the least severe cases may not have come to the hospital so they may not be included in the study)
in case control studies, what are hospital controls
patients at the same hospital as the case but with a different diagnosis
what are 3 considerations for choice of hospital controls
- the illness in the controls should be unrelated to the exposure of interest in the study
- the illness in the controls should have the same referral pattern as the case
hospital controls tend to avoid recall bias, and are cheaper/easier to recruit
what is Berkson’s bias
when hospital controls in a case control study are used and the disease in controls related to the exposure of interest
late look bias is also known as ___
Neyman bias
in case control studies, what is overmatching
matching cases/controls on factors not thought to be relevant to the exposure-disease relationship
when matching in case control studies, what do we not want to match
the exposure or the disease
how does case crossover work
we take a case and look at a time in the past when they were not a case. we then measure the rate of exposure when they became a case, take another time in the past, and compare
what are randomized control trials used for
to determine whether treatments improve prognosis, cure disease, or prevent disease
what are historical controls
we compare the outcomes of patients provided with treatment in the present with the outcomes with the same diagnosis in the past
what is simple radomization
we have a population of study participants, and 2 arms of the study. we randomize half of the participants to each arm
what is stratified randomization
accounts for the fact that even after randomization, we still may end up with groups nonrepresentative of each other by chance (account for confounders)
what is cluster randomization
sorts based on clusters if we are concerned that the intervetion will spread
when are non-randomized historic controls used
when we have a small case number with a high case fatality rate
what is the Hawthorne effect
people change their behavior because they know they are being studied
single blinded studies help control for what
placebo effect
Hawthorne effect
double blinded studies help control for what
observer biases such as the pygmalion effect and researcher expectancy
what is the pygmalion effect
where investigators communicate to or influence patients so that the patients end up having the outcomes they are supposed to have
what is a crossover trial
when participants start in one group and are moved to another treatment group
what is the intention to treat analysis
we analyze the results based on the assigned treatment group
what is the function of factorial design
it allows us to see if two treatments are more effective together than separately
what is a type I error
we conclude treatments are different when in reality they are not (the null hypothesis is true but we thought the alternative hypothesis was true)
what is a type II error
we conclude treatments are not different when in reality they are (the alternative hypothesis is true but we thought the null hypothesis was true)
is type I or II error worse
I
type I- you think you found something but you actually didn’t
type II- you’re looking for something and you can’t find it so you keep looking
how does alpha differ from beta in terms of type I vs type II error
alpha- probability of making a type I error
beta- probability of making a type II error
what does the power of a study tell us
how capable a study is of determining that two treatments differ in their effectiveness when in fact they do differ
what is the equation for power
1-beta
an underpowered study is when the power is __
less than 80%
how does the difference between cure rates in treatment groups help to determine the number of people needed to be recruited to maintain high power
a small the difference between the cure rate, the more people that need to be recruited
what are the main functions in each stage of a clinical trial
I (in patient)- determine maximum tolerated dose, pharmokinetics, and pharmodynamics
II (outpatient)- establish how well the treatment worse and establish dose levels
III (multisite)- spread to diverse populations to generalize results
IV- post-FDA approval, surveillance for adverse effects
in clinical trials, what is the function of the sponsor (holder of drug patent)
write protocol
select doctors to conduct research
determine research questions
who are the primary kind of organization that conducts clinical research trials
contract research organization
when can generic forms of a drug be sold
once the patent expires
what are the 4 requirements for a generic drug to be produced
same active ingredients
inactive ingredients can be different but must be safe
same dosage form (tablet vs capsule, etc)
same route of administration
same pharmokinetics
what is a nested case control study
you take a sample of those with disease and another sample from the population of those who have the disease and compare past exposure