EBM Flashcards
how many phases of drug development are there? what are they?
(I): 3-30, safety, dose-finding, volunteers
(II): 30-50, safety, how effective, how often
(III): 100+, efficacy, safety
(IV): 1000+, post approvalin large population
what are the 5 different randomised control trial designs?
parallel group RCTs factorial design crossover cluser randomisation stepped-wedge
stratification in RCTs
eligible population separated into groups or strata, effectively a separate randomisation within each stratum
only stratify on factors likely to have an impact on outcome (eg tumour site/stage)
most useful in small trials - cluster randomised trials
randomised factorial design
patients are randomised more than once in factorial design
all possible combinations are studied
–> simple 2 way factorial has 4 combinations
pros + cons of factorial design
pros = efficient - 2 trials for price of 1
cons = relies heavily upon the assumption of no interaction between treatments
crossover RCT design
each patient receives ALL treatments
order of receipt is randomised
requires wash-out period between treatments
*only applicable to transient effects in chorinic conditions - diabetes, asthma
pros + cons of crossover trial
pros = smaller sample size needed compared with parallel degisn, patient characteristics remain same
cons = wash out period, not suitbale for intervention that “cures” (used in chronic pain, diabetes)
when would you use cluster randomisation?
intervention at the group level
sometimes only practical design
can reduce contamination
easier to implement intervention to a cluster
implications of randomisation by clusters/practices
sample size needs to be inflated - subjects within practice more alike than subjects in different so independence assumption of statistical tests is incorrect
ethical approval required from both patient + practice
stepped wedge RCT design
order in which unit receives intervention randomised - related to crossover design
used with cluster RCT where efficacy already fairly established
whats the difference between intention-to-treat vs per-protocol analysis?
ITT = more pragmatic, fair comparison, reflect real world, essential for trial integrity
–> analysed in group they were randomised in, irrespective of supsequent changes
PP = more explanatory, NOT reflect real workd
–> analysed in treatment group they actually received
cohort design
a group or population followed over fixed length of time, enrolled at a baseline with comprehensive measurement
all heath events recorded
**very expensive
what is a confounder?
a variable related to the exposure + the outcome but is NOT an intermediary variable between exposure + outcome
issue in cohort studies
how do you calculate the relative risk (RR)
risk of disease in exposed / risk of disease in unexposed
interpreting relative risk
RR > 1 = exposure is harmful
RR < 1 = exposure is protective
RR = 3 = 3x increase risk of outcome in exposed RR = 1.7 = 70% increase RR = 0.8 = 20% decrease
what are the different types of cohort studies?
prospective = cohort followed forward in time from present
retrospective = cohort formed in the past and followed up to the present
pros + cons of prospective cohort studies
defined cohort - detailed exposure records with standardised measurement
time consuming
very expensive
pros + cons of retrospective cohort studies
faster answers
don’t have to employ people to conduct regular follow-up (cheaper)
since exposure is determined after the fact, quality of records has to be checked
advantages of cohort studies
temporality - exposure precedes outcome
no recall bias
can study multiple outcomes associated with rare exposures
can measure incidence
disadvantages of cohort studies
large investment of time, human and financial resources if prospective
large sample sizes
loss to follow-up bias
inefficient for rare diseases
uncontrolled confounding
advantages of case-control study
health events (cases) already defined low investment of time + resources - large databases available
relatively low sample size required
efficient for rare diseases
disadvantages of a case-control study
cannot be sure of temporality
recall bias possible when using questionnaires for eposures
cannot measure incidence
cant measure risk directly
control selction difficult
uncontrolled confounding
cross-sectional study
carried out at a single point in time, so exposure + outcome measured at same time (snapshot in time)
eg census survey
good for estimating point prevalence of disease
cross-sectional study disadvantages
prone to volunteer bias, recall bias and poor response
cannot measure incidence
cannot establish causation