4 - Epidemiological Study Designs Flashcards
Identify the five different epidemiology study designs and what categories they fall under?
What is the methodology of an ecological study?
- Get a group based on their geographical location, e.g by motorway
- Count number of cases of disease you think the location causes, e.g lung cancer
- Gather data on group-level characteristics not individual
EXPOSURE, OUTCOME, GROUP LEVEL
‘hypothesis generating not testing’
What are the different types of measurement in an ecological study?
What are the issues with ecological studies?
- Confounding
- Chance (random error)
- Ecological fallacy (falsely inferring individual value based on group value)
- Definition of characteristics
- Measurement variation
What is the methodology for a cross sectional study?
- Analysis of number of cases at set point in time (prevalence)
- Can assess multiple exposures but as at set point in time can’t see correlation between cases and exposure
How do you generate a sample for cross-sectional survey?
What are the issues with cross sectional survey?
Outline the methodology of case-control study?
- Find cases
- Find non cases (controls)
- Compare & contrast for potential past causal factors (exposures)
- Compare level of exposures in cases and controls
- Always retrospective
What are the issues with case-control studies?
1. Selection bias (controls should reflect study population and be comparable to cases)
2. Information bias - Differential misclassification - information better in one group than the other (e/g can’t remember). Non-differential misclassification - error equal in both groups
3. Confounding
4. Chance (random error)
What is the methodology of a cohort study?
Concurrent or Prospective
- Recruit outcome-free individuals
- Classify into exposed and unexposed groups
- Follow-up each group over time:
I. Count the person-years ‘at risk’ (p-y)
II. Count how many develop outcome (d)
III. Calculate incidence rate (IR = d/p-y)
Historical or Retrospective
- Same but in the past, people aged before got to them
What are the issues with cohort studies?
- Loss to follow up (differential loss and survivor bias)
- Information bias (differential and non-differential misclassification)
- Confounding
- Chance (random error)
What is survivor bias?
- Those who survive or keep in contact tend to be the healthy ones so not true picture of incidence rate
How do you analyse data from each study?
What is the main differences between cohort, case control and cross-sectional studies?
How do you analyse cohort studies?
What are the three ratios used to describe relative risk?
- Rate Ratio
- Odds Ratio
- Risk ratio
How do you analyse case-control studies?
How can you deal with confounding?
- Minimise confounding in selection (matching)
- Adjust for confounding in analysis (adjusting)
Outline the positives and negatives of all study designs
Cohort not good for rare outcomes
