Module 1; Measuring health and dis-ease in populations Flashcards
what is the equation for the frequency of disease?
N/D/T
what is the gate frame made up of?
PECOT (population, exposure, comparison, outcome, time)
does the gate frame require all components of PECOT?
no, not all the time
what is a cohort study?
group of participants that are allocated into several groups (EG/CG) and followed up over a period of time
are cohort studies observational or experimental?
observational, and not random
do cohort studies measure incidence or prevalence
both
what is a cross sectional study
outcome + comparison at the same time, calculates the average
do cross sectional studies measure incidence or prevalence?
prevalence
do incidence studies count categorical or numerical events?
categorical
do prevalence studies count categorical or numerical events?
both
what are some problems with the incidence/prevalence
- change if high incidence, low prevalence (people dying)
- change if low incidence, high prevalence (obesity)
advantages/disadvantages of prevalence
easy + cheap to measure
only numerator + denominator
dirty because of cure rate and health rate
advantages/disadvantages of incidence
difficult and expensive to measure
only determined by a disease risk in a population
number, denominator and time measured
what are RCTS?
random controlled trials
cohort studies, but participants are randomly allocated
are cohort studies ethical?
yes because they are observing, not experimenting
however these are less accurate than experimental
risk difference
ego-cgo
risk ratio
ECO/CGO
differences between risk difference and risk ratio
risk difference
- has units
risk ratio
- has no units
- rr reduction, RR > 1
- rr deduction, RR < 1
overall risk difference > risk ratio because its more accurate.
errors in epidemiological studies
random error (chance)
non random error (systematic error)
what is the acronym for non random errors?
RAMBOMAN
Ramboman
recruitment
- representative of the whole population? selection bias
rAmboman
allocation
- usually by measurement or by random allocation
measurement ; e.g. measurement smokers vs drinkers. questionnaire
random allocation; prevents bias
- reduces effect of counding
- random trials = lots of effort and learning
raMboman
maintenance
long experiment > short experiment.
participants may not stay in allocated groups
participants lost to follow up
not a problem in prevalence bc its a one time measure
ramBoman
double blind vs single blind
double blind prevents doctors from having a trait to distract their feeling
rambOman
objective
- should have no confusion in interpretation + subjective words can be confusing
- ecological studies can struggle w this bc different words are subjective
what are ecological studies
allocation of whole population rather than individuals
e.g. average __ levels
what is a problem with ecological studies + how to combat
confounding!!!
to combat ; use ecological randomised trials.
something special about cohort studies
they can start as cross sectional, and then follow up to become cohort
can cross-sectional studies change?
no they stay cross-sectional
what are the strengths and weakness of ecological studies?
+low random error
+cheap
- confounding common
what are the strengths and weaknesses of rcts?
-difficult to to have representative sample, population are often highly motivated volunteers
+random allocation!! -confounding
-unethical to subject ppl to harm
-maintenance error common
-expensive + small
strengths and weaknesses of cross sectional
+easier to recruit
-confounding common
+ethical
+maintenance error not an issue
+less expensive
-only measure prevalance
–reverse causality (which came first?? exposure/outcome)
strengths and weaknesses of cohort studies
+easier to recruit rep pop
- more confounding errors
+ethical to study harm
-maintenance error common
+less expensive than rct
+can measure incidence + prevalence
how to choose bt study designs
depends on question + what is possible (ethicality, practicality, affordability)
what is a meta-analysis
mathematical combination of studies bc they are too small by themselves
strengths + weaknesses of meta analyses
low random error +
depends on quality of steps
what are the two types of random error?
random measurement error (human error)
random sampling error (not representative of the whole population
95% confidence interval
there is about a 95% chance that a true value in a population lies within the 95% confidence interval
in 100 identical studies using samples from the same population, 95/100 of the 95% cis will include the true value
what is statistically significant?
if ego/cgo dont overlap OR the 95% rr/rd does not overlap the no error line (which is when cgo=ego)
statistically insignificant
when ego/cgo overlap
or rr/rd overlaps with the no error line