Module 1 Flashcards
Numerator
Number of cases of disease
Denominator
Number of people in a population
Frequency of disease
(N/D)/Time
Population
A group of people who share one or more common feature
Epidemiology
The study of occurrence and causes of disease in groups or populations at one point or over a period of time
PECOT
Population Exposure group (E.G.) Comparison Group (CG) Outcome Timing
EGO
Exposure group occurrence
EGO=a/E.G.
CGO
Comparison Group Occurrence
CGO=b/CG
Incidence
Number of disease events occurring over a period of time
EGO & CGO are called incidence of occurrence
Prevalence
The number of people with a disease who are counted at one point in time
EGO & CGO are called prevalence of occurrence
Numerical Outcomes
A quantifiable value e.g. Heart rate per minute
Categorical Outcome
Specific set of options e.g. Yes/no
Period Prevalence
We measure if the event has occurred during a period and not the number of occurrences
Eg. If an asthma attack has happened in the past year
Point Prevalence
Measure Prevalence of something at one point right now
Eg. Level of nicotine in blood at one point
Ecological Study
Comparing groups of populations rather than groups of individuals
Can be RCT, cross sectional, cohort
Confounding is common
Risk Difference - Absolute Inequalities
EGO-CGO
IF RD = 0 then treatment has no effect
Risk Ratio
EGO/CGO
If RR = 0 effect of treatments are the same
Relative Risk Reduction
If RR is less than 1
Relative Risk Increase
If RR is more than 1
R (RAMBOMAN)
Recruitment
Are participants a representative sample from the known population?
A (RAMBOMAN)
Allocation
Was allocations accurate?
Is confounding involved?
Confounding
Difference between EGO and CGO apart from the main difference being studied
To deal with confounding we divide the study into sub studies so participants with the confounded are all in one study
M (RAMBOMAN)
Maintenance
If possible we randomise the population otherwise we ask about disease
Need to maintain initial exposure/comparison groups
B (RAMBOMAN)
Blind
Unbound study: investigators and patients know who is taking treatment or placebo
Single blind: investigators know, patients do not
Double blind: neither investigator nor patient know
O (RAMBOMAN)
Objective
Objective outcome as it is easiest to measure
E.g. Dead or alive
M (RAMBOMAN)
Measurement of Outcomes
Everyone has different interpretations of subjective outcomes
- Recall Bias
Age standardisation
(Disease/study population) x age standardised population
(Total of ages of age specific death rates/ age standardised population) x age standardised disease rates
Random Error
Extreme events are often chance restarting of these will often produce less extreme results - regression to the mean
Random Measurement Error
Identical measures of exposure/outcome in similar people can result in different outcomes over time
Random Sampling Error
We can’t study everyone only a sample of the population
95% Confidence Interval
A measure of the amount of random measurement/sampling error when you’ve only done one study
Reduces random error
Cohort Study
Study of individuals allocated to groups over time
Observational longitudinal study
Involving incidence
Used for investigating risk factors for disease
Main errors: how accurate is allocation, maintenance error
Cross Sectional Study
Study of individuals allocated to groups at the same time
Involving Prevalence
No maintenance error
Main error: confounding, reverse causality
Randomised control Trial
Study of individuals randomly allocated to groups and counted over time
Involves incidence (sometimes Prevalence)
Commonly used for studying effects of drugs - reduces confounding
Main errors: having motivated volunteers, random error common
Meta Analysis
A study of studies
Combining the results of a number of small studies is similar to conducting a larger study and it reduces the amount of random error
Only as good as the studies included
No effect line: if studies cross this are less statistically significant