Module 1 Flashcards
what does GATE stand for
Graphic Approach to epidemiology
PECOT
P - population of participants
E - exposure groups
C - comparison group
O - outcomes
T - time
CGO
Comparison group occurrence B/CG/t
EGO
exposure group occurrence A/EG/t
Features of a Cohort study
- long-term follow-up
- allocation by observation
- time going down
- common confounding
- can measure incidents and prevalence
Incidence
Measure of the onset of a disease. e.g deaths, may be hard to measure
Prevalence
measure how much disease is in the population
What is an issue with measuring prevalence
That prevalence is a dirty measure as it has the cure and death rate involved with the calculation. not good for temporary events e.g asthma attack
What is one way to measure prevalence
Go back in time e.g during the past year have you xyz
RCT
Randomized control trial - best if ethical and practical
Why cannot do an RCT
if the intervention may be harmful, we cannot get people to do it as it may compromise health. only if the potential benefit to health.
How is an RCT conducted
By random allocation of individuals or groups
What is ethical around harmful exposures
It is not ethical to intervene but it is ethical to observe
Why are long term RCT hard to do
Many people do not remain in their groups long-term
Ethical errors
often hard to recruit pop that are representative
RD
EGO - CGO
It is the absolute difference
Has units
RR
EGO/CGO
No units
Relative difference
No Effect line RD
0
No Effect line RR
1
RRI
relative risk increase
If RR is over 1 can go RR/1 to get percentage increase
RRD
If RR less than 1. Can go 1-RR/1 to get percentage decrease
Maintenance Error
Loss to follow up
Did they change groups
Blind or Objective measures
Subjective vs Objective
Double Blind
How were outcomes measured
Ecological studies
Allocation of populations to EG or CG
very common confounding
cheap to do
can be cross sect, cohort or RCT
Cross Sectional
Measure done at one point in time
reverse causality
Cheaper
Measuring prevalence
Meta Analysis
Can combine multiple studies to one large study. increases sample size and decreases the random error
narrow definition of health
the absence of death, disease or disability
broad definition of health
the capacity to do what matters most to you
study validity
a study with only a small amount of random or non-random error is considered to be a valid study
recruitment error
are the participants a representative sample from a known population?
layers of the triangle
setting, eligible population, actual participants
confounding
when the exposure is mixed with another factor that is also associated with the outcome
stratified analysis
dividing the study into sub-studies (strata) so participants with the confounder are all in one sub-study
regression to the mean
repeating measurements or studies with extreme results multiple times usually gives less extreme results
random sampling errors
the smaller the sample, the greater the chance the sample will be different from the whole population
95% CI
there is about a 95% chance that the true value in a population lies within the 95% confidence interval
bradford hill criteria: [7]
- temporality
- strength of association
- reversibility
- biological gradient (dose-response)
- biological plausibility
- consistency of association
- specificity
temporality (BH)
- first the cause then the disease
- essential to establish a causal relation
strength of association (BH)
the stronger the association, the more likely to be causal in absence of known biases
reversibility (BH)
the demonstration that under controlled conditions, a change in exposure results in a change in the outcome
biological gradient (BH)
incremental change in disease rates in conjunction with corresponding changes in exposure
biological plausibility of association (BH)
does the association make sense biologically?
consistency of association (BH)
replication of the findings by different investigators, at different times, in different places, with different methods
specificity of association
a cause leads to a single effect and an effect has a single cause
the epidemiological triad
host, environment and agent
a cause of a disease
an event, condition, characteristic (or combination of these factors) which play an essential role in producing the disease
sufficient cause
the whole pie
a minimum set of conditions - without any one component disease would not occur
component cause
each factor/slice is a component cause
necessary cause
a factor (or component cause) that must be present for a specific dis-ease to occur
problems with the causal pie model
assumes all causes are deterministic and fails to capture dose-response relations as a continuum
