EBM 2 - EPIDEMEOLOGICAL STUDIES Flashcards
5 STUDIES
2 CLASSIFICTION in each study
in order of strongest to weakest causality ;
- RCT, Cohort, case control, cross sectional & ecological
- observational & interventional
RCT- how to plan one
how to conduct one
do participants that do not take the drugs they are supposed to or dropout – are they scratched ?
PLAN;
- null & alternative hypotheses
- eligibility of candidates
- define intervention and control group within the candidates
- calculate sample size
- specify outcome measures ( e.g. mortality what are we going to compare the results to?)
CONDUCT;
- get ethical approval
- randomisation process
- consent form candidates
- allocation CONCEALMENT
- blinding
all candidates regardless of how well they participate in the trial are included to AVOID BIAS
CLINICAL EQUIPOISE
- Equlibrium
- balancing harm or benefits of treatments and placebos
- only ethical to give x to control group if its not harmful etc
allocation concealment vs blinding
when might blinding be impossible?
- allocation sequence (randomisation process) is HIDDEN from clinicians trying to recruit
- blinding; patients & clinicians are unaware of what treatment allocation (single or double blinded trials)
during SURGICAL RCT’s
RCT results - how we calculate results
what do we use?
INTERPRET what these values could mean
RR; RISK RATIO; relative risk aka
risk of outcome in exposed or treatment grp divided by;
risk of outcome in non exposed or control group
NOTE ; arm refers to groups
if; RR is more than 1, it suggests that the exposure predisposes the disease or outcome
RR less than 1; exposure protects vs the disease.
RR measures association of exposure with outcome but this relationship could be CAUSAL !!!
RD; RISK DIFFERENCE;
risk of outcome in exposed - risk of outcome in non exposed
think; RD is the EXCESS risk involved due to exposure
good and the bad - RCT
GOOD;
- strongest evidence of causality between intervention and outcome
- low levels of CONFOUNDING (CH4)
BAD
- not always ethical
- high dropout rate if results or side effects not desirable
relative risk reduction
NNTB + NNTH means ;
formula
RRD; 1 - RR
EG ans is 1-.29 =.71 = 71% reduced risk of disease with treatment
but clinically we use the following ;
No. Needed to Treat to Benefit/Harm formula; invert RD; 1 over; RD
NNTB; round up to nearest integer
NNTH; round down to nearest integer
explanation ; e.g. and is 15
need to treat 15 patients to prevent one patient getting a particular disease.
more meaningful
COHORT design study aka longitudinal studies
- what is it?
- different types
- observational study
- exposure of interest is measured at start of study amongst people who are free of disease
- participants are the followed up to see if they get the disease at the same rate as people who were not exposed
2 TYPES ;
PROPSPECTIVE cohort study;
- the exposure is measured in the present day and participants or COHORT are followed to see if they develop the disease in the future
present- future
HISTORICAL cohort study;
- exposure was measured in past from historical records usually and the outcome can be anything from the past up to the present day
past- present
Q’s;
exposed; YES or NO
disease; YES or NO
Thus there are 4 options (box diagram)
How to calculate RR ?
formula;
what does it mean again?
explanation using figures
how to calculate risk difference
units?
BOX method;
DISEASE?
YES NO TOTAL
- exposed? YES d1 h1 n1= D1+h1
NO d0 h0 n0= d0 +h0
TOTAL
SUBSCRIPTS; 1; exposed 0;non exposed d; disease h; healthy
after this table whats the formula
d1/n1 over;
d0/n0
means ; risk of disease in the exposed divide by;
risk of disease in the non exposed
e.g. if RR is 3- 3 times more likely to het the disease if u are exposed than if you are not exposed
RD;
what? ; excess risk of disease due to exposure than if u were non exposed
formula;
d1/n1 – d0/n0
RR ; no units
RD ; UNITS ! usually per 1 thousand etc
good and the bad ; cohort studies
good;
- minimises chance of bias as exposure measured b4 outcome in prospective cohort study
- allows testing for multiple outcome hypotheses for a particular exposure
bad;
- difficult to measure exposure accurately in historical CS if records are not accurate
- problem with CONFOUNDING ? learn in ch4
- time consuming
CASE CONTROL STUDIES
what are they
- cases ; ppl have the disease
- controls ; no disease
SO;
- basically choose people to be in both groups (case&control)
- estimate exposure (retrospectively)
if exposure is more common among cases its suggest the exposure predisposes the outcome
if exposure is less common in cases it suggests it PROTECTS vs the disease
how do we distinguish between case control and historical cohort study
simple terms ;
- DURATION of study ; CC usually take place over a shorter time period than HC (because we already have outcome)
- START POINT; CC starts with ppl w & w/0 outcome /disease, HC starts with ppl who do not have the disease and is defined by whether or not they had the exposure of interest
- HC will include dead ppl as exposure and outcome records are already available
NB; incidence rates &; RR can only be calculated in the cohort studies
not in CC as groups have been chosen so that the same no. of people in each so no ratios or rates could be determined
ODDS ratio
- what does it represent or mean
- how to calculate and formula
why choose case control?
means; risk of disease among exposed over non exposed
calculate;
set up box method in same way for RR & RD
Formula;
d1/d0over;
h1/h0
why CC?
most cost efficient when disease is rare
good & bad - CASE CONTROL
Good;
- cost efficient study for diseases that are rare
- relatively quick as investigator does not need to wait for cases to appear i.e. they have already happened
- allows testing of multiple EXPOSURE for a particular disease
bad;
- RECALL bias form participants
- which came first tho - the exposure or the disease = reverse causality
