introduction to epidemiologypublicifsv.sund.ku.dk/~pka/epi19/ksl1.pdf · 2019. 1. 25. ·...
TRANSCRIPT
Katrine Strandberg-LarsenSection of Epidemiology January 29th 2019
Course: ‘Epidemiological methods in medical research’
Introduction to epidemiology
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Who are you?
Introduce yourselves in 2 minutes:
- Name
- Workplace
- Aim of you PhD – specify main epidemiological design
- Statistical software
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
What is epidemiology?
Etymology
• Epi = on, with
• Demos = people, population
• Logos = knowledge of
• Epidemiology = The knowledge of what lies on people
Purpose
• Monitor health and illness in populations – determine burden of disease
• Identification of cause and risk factors of disease
• Study natural history and prognosis of disease
• Evaluate preventive and therapeutic measures
• Provide evidence for public health policy and clinical practice
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
The unit of interest
Descriptive epidemiology
Department of Public
Health
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Increased lung cancer mortality
• Steep and continuous increase in lung cancer mortality during the first part of the 20th century in Western countries.
• For example the lung cancer mortality of English men increased from 1,1 pr. 100.000 in 1901-20 to 10,6 pr. 100,000 in 1936-9.
• In UK the lung cancer mortality increased 6 times for men and 3 times for women from 1921-30 to 1940-44.
Socialmedi
cin
Department of Public
Health
Department of Public
Health
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
The role of epidemiology
Is to assembling a diverse body of facts from different sources into a coherent explanation.
”Epidemiology is something more than the total of its established facts. It includes their orderly arrangement into chains of inference which extend more or less beyond the bonds of direct observation”
American epidemiologist Wade Hampton Frost
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Assembling and systemizing evidence
Descriptive Analytical
Analytical epidemiology
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
What causes of disease?
Socialmedi
cin
CauseExposure
X
ConsequenceOutcome
Y
EffectAssociation
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Different words, same thing
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
1968 data on cigarette consumption and lung cancer
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Ecological/correlation studies
• Make use of available aggregated data
• Plot outcome frequencies against exposure frequencies in e.g. countries or states
Department of Public Health
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Are immigrants more likely to be well-off?
Department of Public Health
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Immigrant status and income: individual level data
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
The ecologic fallacy
Bias that may occur because an association observed between variables on an aggregate level does not necessarily represent the association that exists at an individual level
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Patients with lung cancer
Socialmedi
cin
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Comparisons of cases and controls
Socialmedi
cin
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Case-control design
In brief:
Collect data on exposure on a group of cases and a group of controls
Compare the exposure frequency in cases and controls
The selection of controls must not depend upon exposure
Socialmedi
cin
Cases
Controls
ExposedUnexposed
ExposedUnexposed
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
What did they observe?
Socialmedi
cin
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Conclusion
Socialmedi
cin
Doll and Hill. BMJ 1950
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Tried a new approach
Socialmedi
cin
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Cohort design
Socialmedi
cin
Population
Exposed
Unexposed
Incident case
Number at risk/risk time
Incident case
Number at risk/risk time
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
What they observed after 5 years?
Socialmedi
cin
Doll and Hill. BMJ 1956
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
What they observed after 10 years?
Socialmedi
cin
Doll and Hill. BMJ 1964
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Two variables are associated if…
1. Causal effect
A Y
2. Share a common cause
C
A Y
3. Condition on a common effect
A Y
Socialmedi
cin
S
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Type of biases
1. Causal effect
A Y Reverse causation
2. Share a common cause
C Confounding
A Y
3. Condition on a common effect
A Y Selection bias
Socialmedi
cin
S
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Randomization
Can we use this design to:
Smoking Lung cancer
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Randomization without intervening
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Genes are used as IV
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
CHRNA3 is associated with smoking
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Confounders according to smoking and genotype
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Smoking and all-cause mortality
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Genotype and all-cause mortality
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Mendelian randomization analyses
punktopstilling, brug
”Sidehoved / Sidefod”.Indføj ”Sted og dato” i
Assembling and systemizing evidence
Descriptive Analytical