Epidemiology, Gate frames & Incidence/Prevalence Flashcards
What is the goal of epidemiology & basic equation used
Studying disease occurance in a population over time
(no. of ppl with disease) / (No. of ppl in the population)
What does numerator and denominator represent
Numerator is those with disease and prevalence is the study population sharing similar characteristics
Components of the Gate frame
Triangle - study population
Circle - EG and CG
Square - EGO and CGO
O - Outcome in square
Horizontal - One time measurement
Vertical arrow - Overtime measurement
What do you do when a numerical measure isn’tn converted into a categorical?
Calcultae the mean or median excluding c or d (No disease people from both EG and CG) as this assumes everyone has a disease
EGO and CGO acronym meaning
Exposure group occurance and Comparison group occurance
Incidence components
Easy to observe disease events measured over time via categorical measurments e.g. death
Clean measure but hard to measure as observation occurs overtime
Prevalence components
Disease occurance measured at one point in time when transition from non diseased to diseased is blurry viz categorical and numerical measures
Dirty measure (Affected by incidence, death and cure)
Easy to measure as it happens at one point in time
Prevalence pool
Leftover drizzle from incidence and size depends on drizzle rate and loss to death and cure
Prevalence pool
Leftover drizzle from incidence and size depends on drizzle rate and loss to death and cure
Direction of arrows for one point in time measurement
Horizontal
Direction of arrow for longitudinal study
Vertical
Direction of arrow for timeframe from the past - Prevalence study
Top left corner
Association (Comparing disease measures)
Intitialdisease occurance should be estimates of association
Relative risk
Ratio of occurance with no units
Relative risk = 1
No effect value - there’s no difference in the effect of E and C on outcome
Equation for Relative risk > 1
Relative risk increase = (RR -1) x 100
Equation for relative risk < 1
Relative risk reduction = (1-RR) x 100
Absolute risk difference
Difference in disease occurance with units and contains more information than RR
Absolute risk difference = zero
No risk difference thus no difference between EGO and CGO
Absolute risk difference < 0
Ideal risk difference
Absolute risk reduction
Absolute risk is lower in EG
Absolute risk increase
Absolute risk is greater in EG
Randomised control trial key components
Study the effects of ethical and practical treatments
by random allocation into EG and CG via experiments
Rarely used to discover risk factors instead more in placebo studies
Long term RCT
Harmful treatment can be used as the test is now only observational but difficult to do well due to maintenance
Randomised control trial strengths
- minimise cofounding
- measures incidence and prevalence
- ensures participants in EG and CG are similar
Randomised control trial weaknesses
- Hard to recruit representative eager participants
- A-lot of maintenance error
- Very expensive and too small
- Too small produce insufficient results due to expensive conduct
Non random error
Errors in study design, poor ramboman
Recruitment error
Study population representativness/elgibility - May be over or under representation
Selection bias
Recruitment error
EG is recruited from a very different group to CG
Cofounding error
Recruitment error
Studies exposures mixed with external factors effecting outcome causing a bias
Non response bias
Recruitment error
Large proportion of eligible population don’t respond and if 70% of the responders are significantly different to population
Allocation error
How participants were allocated into EG and CG - measurment or randomly
Measurement error
Allocation error
EG and CG is measured incorrectly leading to wrong allocation
Allocation by measurment
Participants are measured objectively or subjectively to be allocated
Maintenance error
Mainitaining participants in alloacted groups - occurs more often in long term studies as patients can be lost to cure, death, leaving study etc
Compliance
Maintenance error
Percentage of participants remaining in allocated group
Contamination
Maintenance error
Percentage of participants crossing over allocatedn groups
Co- intervention
Maintenance error
Did EG and CG recieve unequal interventions in follow up
Blinding
Reduces cofounding by preventing personal interpretation - single or double
Objective outcomes don’t need blinding
Adjustment
Dividie participants into strata and compare differing groups to reduce cofounding
Random allocation
randomly allocate into EG and CG
Random error
Biological variation causes participants to be moving targets
Random sampling error
Smaller sample the less representative of population
Every sample is the best estimate of the truth of a population
Where does random error occur
EGO, CGO , RR and RD
95% confidence interval
Measure of the range and size of random error
95% chance the true value of population lies within the CI
The wider the interval the more random error = 99% CI
Statistical signifiance
When EGO and CGO confidence intervals dont overlap the no effect line
RD doesn’t cross no effect line
RD is statisticaly significant
RD overlaps no effect line
No statsitical significance in EGO AND CGO usually due too to much random error to determine a real difference
RR overlaps 1
No statistical signficance
Reducing random error
Get bigger samples to narrow the Confidence interval
Regression to the mean
Extreme events are often due to chance thus repeating them eventually gives less extreme results closer to the mean
Meta anaysis
Combine many small good studies to form one big meta study - Used especially in RCT when too much random error blurs effects of study
Cohort study
For unethical and unpractical studies done overtime to measure incidence and prevalence
Cohort study weakness
Cofounding and maintenance error
Cohort study strengths
- Easier to recruit representative sample
- Cheaper, larger and easier
- Little measurment and random error
Cross sectional study
Exposures and outcome measured at once for prevalence observational study
Cross sectional study strengths
- Easiest to recruit representative sample
- More cheap and large
- Ethical to study harmful treatment
- No maintenenance error
Cross sectional study weaknesses
- Cfounding common
- Reverse causality - Did outcome or exposure come first
Ecological study
Participants are countries thus many EG
DOne via surveys usually
Longitudinal or cross sectional
Ecological study strengths
- Large thus low random error
- Very cheap - Based on already published study
Ecological study weaknesses
- Often unadjustable cofounding
Descriptive study
Dsecribes frequency of health behaviour, risks and outcomes
Analytical study
Analyses descriptive study results
Measures incidence AND prevalence
RCT and cohort studies
Only measures prevalence
Cross sectional studies
Individual participation studies
- RCT
- Cohort
- Cross sectional
Ecological studies
- RCT
- Cohort
- Cross sectional
SImilarties in study designs
- GATE and PECOT
- Rambom
- Random error
Differences in study desgins
- Not all involve EGO and CGO calculation
