Epidemiology
Lecture 1 – Introduction & study designs
Epidemiology: Study what is upon people.
Richard Doll First epidemiologist, combined medicines with mathematics.
Studied the enormous rise of lung cancer after 1900 tried to find relation with
the environment (e.g. the new material tarmac, motor car fumes, smoking,
indoor cooking etc.).
Important epidemiologic terms:
Determinant is the same as independent variable exposure cause
Outcome is the same as dependent variable effect
Descriptive epidemiology: Study of disease patterns.
What? Who? Where? Why? When?
What is the frequency of Chlamydia among students?
Analytic epidemiology: Study of disease determinants
“It is all about comparing persons, groups, etc.”
Are epidemiology students more likely to contract Chlamydia than theology
students?
Types:
o Etiology = the cause of disease
o Prognostics = course of disease
o Diagnostics = association with disease
Observational: Observe and measure a set of characteristics in a defined sample
population. Subgroups:
Cohort study: Study population comprises a group of persons that share at
least once characteristic. Further divided in:
o Retrospective: Outcome assessed at present and determinant asked
from the past
Do students who are suffering from autism nowadays, ever
consumed stevia?
Could calculate prevalence and do odds ratio measurement of
association
o Cross-sectional: Determinant and outcome assessed at the same
point in time
Do student who are suffering from autism nowadays,
consuming stevia at this moment?
Could calculate prevalence and do odds ratio measurement of
association
o Prospective: Determinant assessed at present and outcome
assessed in the future
Do students who consume stevia nowadays, develop autism
in future?
Important to start with people who do not already have the
outcome (or are at risk, will mess up the study)
, Could calculate prevalence, cumulative incidence and
incidence density and do odds ratio measurement of
association and RD/RR
Case-control study: Select patients based on outcome and find comparable
controls
o Patients with lung cancer + patients with brain tumor in the same
hospital
o Always retrospective
o Important for controls: represent persons who would have been a
case, would they have the disease of interest? And, represent for the
exposure to the determinant in the study: selection of controls
should be independent of exposure.
o Could calculate odds ratio measurement of association
Experimental: Manipulate a characteristic within the defined sample population
with the purpose of studying the effect this has.
Cohort study (prospective)
o New drug vs old drug, drug vs placebo etc.
o Patients are allocated to the determinant (intervention)
o An experimental study is always a prospective design
o Could calculate prevalence, cumulative incidence and incidence
density and odds ratio measurement of association and RD/RR.
Once you are in a cohort, you are never able to leave.
Ecological study: Subtype, e.g. comparing among countries.
Unit of observation is a group of persons
Usually registries are used to collect data
Design can only be used to generate hypotheses
Two types:
o Geographical correlation studies
Different geographical regions are compared with respect to
some exposure and some disease/outcome
Example: Two different countires are compared with respect
to the mean alcohol consumption and heart-disease related
mortality
o Time trend studies
Within one region, different time periods are compared with
respect to some exposure and some disease/outcome
Example: In a given country, two different time periods are
compared with respect to the mean alcohol consumption and
heart-disease related mortality
Ecological fallacy = determinant and outcome are being assessed at
aggregated (group) level and from different registries. Causal inferences at the
individual level are not permitted! Doing so is referred to as the ecological fallacy.
Pro’s and con’s of different cohort studies:
Retrospective Cross-sectional cohort Prospective cohort
studies studies
Inexpensive Selection Inexpensive Cause not Limited Expensive
bias before chance of
outcome: selection
Causality bias:
Lecture 1 – Introduction & study designs
Epidemiology: Study what is upon people.
Richard Doll First epidemiologist, combined medicines with mathematics.
Studied the enormous rise of lung cancer after 1900 tried to find relation with
the environment (e.g. the new material tarmac, motor car fumes, smoking,
indoor cooking etc.).
Important epidemiologic terms:
Determinant is the same as independent variable exposure cause
Outcome is the same as dependent variable effect
Descriptive epidemiology: Study of disease patterns.
What? Who? Where? Why? When?
What is the frequency of Chlamydia among students?
Analytic epidemiology: Study of disease determinants
“It is all about comparing persons, groups, etc.”
Are epidemiology students more likely to contract Chlamydia than theology
students?
Types:
o Etiology = the cause of disease
o Prognostics = course of disease
o Diagnostics = association with disease
Observational: Observe and measure a set of characteristics in a defined sample
population. Subgroups:
Cohort study: Study population comprises a group of persons that share at
least once characteristic. Further divided in:
o Retrospective: Outcome assessed at present and determinant asked
from the past
Do students who are suffering from autism nowadays, ever
consumed stevia?
Could calculate prevalence and do odds ratio measurement of
association
o Cross-sectional: Determinant and outcome assessed at the same
point in time
Do student who are suffering from autism nowadays,
consuming stevia at this moment?
Could calculate prevalence and do odds ratio measurement of
association
o Prospective: Determinant assessed at present and outcome
assessed in the future
Do students who consume stevia nowadays, develop autism
in future?
Important to start with people who do not already have the
outcome (or are at risk, will mess up the study)
, Could calculate prevalence, cumulative incidence and
incidence density and do odds ratio measurement of
association and RD/RR
Case-control study: Select patients based on outcome and find comparable
controls
o Patients with lung cancer + patients with brain tumor in the same
hospital
o Always retrospective
o Important for controls: represent persons who would have been a
case, would they have the disease of interest? And, represent for the
exposure to the determinant in the study: selection of controls
should be independent of exposure.
o Could calculate odds ratio measurement of association
Experimental: Manipulate a characteristic within the defined sample population
with the purpose of studying the effect this has.
Cohort study (prospective)
o New drug vs old drug, drug vs placebo etc.
o Patients are allocated to the determinant (intervention)
o An experimental study is always a prospective design
o Could calculate prevalence, cumulative incidence and incidence
density and odds ratio measurement of association and RD/RR.
Once you are in a cohort, you are never able to leave.
Ecological study: Subtype, e.g. comparing among countries.
Unit of observation is a group of persons
Usually registries are used to collect data
Design can only be used to generate hypotheses
Two types:
o Geographical correlation studies
Different geographical regions are compared with respect to
some exposure and some disease/outcome
Example: Two different countires are compared with respect
to the mean alcohol consumption and heart-disease related
mortality
o Time trend studies
Within one region, different time periods are compared with
respect to some exposure and some disease/outcome
Example: In a given country, two different time periods are
compared with respect to the mean alcohol consumption and
heart-disease related mortality
Ecological fallacy = determinant and outcome are being assessed at
aggregated (group) level and from different registries. Causal inferences at the
individual level are not permitted! Doing so is referred to as the ecological fallacy.
Pro’s and con’s of different cohort studies:
Retrospective Cross-sectional cohort Prospective cohort
studies studies
Inexpensive Selection Inexpensive Cause not Limited Expensive
bias before chance of
outcome: selection
Causality bias:
