Behavioural science Flashcards
collects data from a group to assess frequency of disease
cross-sectional study
compares groups with and without disease
cae-control
looks for prior exposure or risk factor
case-control
compares a group with a given exposure to those without
cohort study
ORs come from
case-control
RRs come from
cohort
phase I clinical trial
is it safe?
small number of healthy volunteers
phase II clinical trial
does it work?
small number of diseased patients
phase III clinical trial
is it as good or better?
large number of patients RTC
phase IV clinical trial
can it stay?
surveillance of patients after treatment
Sensitivity of a test =
= true positives / (true positives + false negatives)
Specificity of a test =
= true negatives/ (true negatives + false positives)
description sensitivity
probability that when the disease is present the test is positive
description specificity
probability that when the disease is absent the test is negative
positive predictive value =
TP / total positives
describe positive predictive value
proportion of positive results that are true
incidence rate =
number of new cases/ number at risk
over a specific time period
prevalence =
number of existing cases/ total in population
at a certain time
when is prevalence greater than incidence?
chronic disease
Risk ratio =
(person with disease + risk factor / all those with risk factors )
minus
(person with disease and no risk factor/ all persons without risk factor)
odds ratio =
(person with RF and disease x person without RF or disease) /
(person with risk factor and no disease x person with no risk factor and disease)
atrributable risk
proportion of disease occuring that are attributable to exposure
attributable risk =
(RF+disease / RF) - (noRF+disease / no RF)
relative risk reduction =
1 - RR
absolute risk reduction =
actual difference in risk
NNT =
1/ ARR
NNH =
1/ AR
Berkson bias
study population comes from hospital and is therefore less healthy than the general population
Healthy worker effect
study population is healthier than the general population
non-response bias
differential response
procedure bias
information gatsubjects in different groups are not treated the same
pygmalion effect
observer-expectancy bias
thinks there should be an effect, so looks harder for one
confounding bias
related to both exposure and outcome, but not on causal pathway
crossover study
subjects act as their own controls
lead-time bias
early detection is confused with increased survival
mode =
most common
SEM =
SD / sqrt n
95% =
2 SDs