Study Guide Epidemiology and Global Health
Problem 1 – Study Designs
Study design: plan of how to conduct a study to answer a hypothesis
Observational study (measured by observation, not intervened)
- Descriptive (describe the occurrence of disease)
o Used to make hypotheses
- Analytical (analyses the relationships between exposure and disease)
o Ecological (population)
Correlational
Generate hypotheses
Unit: groups of people
Advantages: use populations with differing characteristics, and
existing data can be used
Test for potential confounders
o Cross-sectional (individual)
Measure of exposure at a given point of time
Key question: does the exposure precede or follow the effect?
For fixed exposures (ethnicity, blood type)
Advantages: the timing of the exposure is known (whether it
happened before or after the exposure), and looking at the
possibility causes of disease can make the study successful
If exposure is not measured before the disease the study gives
disease prevalence
Recall bias is high (people might not remember how much they
smoke for example)
Trends can be measured if people are examined in several
points in time
o Case control
On people that have the disease
Backtrack what leads to the disease
Compared to people without the disease to see normal
occurrence with similar characteristics
Disadvantage: can be biased as the past is examined, and it is
hard to find control groups that are very similar
Relative risk is calculated (compared to controls)
o Nested case control studies
Using cases from cohorts for which exposure information is
already available (checking whether they had the disease or
not)
Cohort is already designed
o Cohort
Aka follow-up study
Start with healthy people (see whether they get the disease
during the follow-up)
Can be pro and retrospective study
(Difference from case control is that the subjects are healthy to
start out with)
Advantages: insight into pattern of exposure and the disease,
and insight into causality
, Disadvantages: expensive, and large group is required
Experimental study (assign exposures, manipulate certain factors and measure the
effect)
- Randomized control trial
o Randomly selected people from two groups and are compared
o E.g. vaccine studies
- Field trial
o On healthy, but presumed to be at risk
o Data collection takes place in the field
o E.g. lead levels in children for those that were raised under lead
regulated environments compared to those that were not
- Community trials
o Do not focus on the individual
o Good to measure disease that affect the community (eg
cardiovascular disease)
o See the effects of social conditions on disease development
Sources of error:
- Random error (eg variation in people and by chance in small sample)
- Systematic error
o Selection bias
o Measurement error
Recall bias
Observer bias
o Response bias
- Confounding
Problem 1 – Study Designs
Study design: plan of how to conduct a study to answer a hypothesis
Observational study (measured by observation, not intervened)
- Descriptive (describe the occurrence of disease)
o Used to make hypotheses
- Analytical (analyses the relationships between exposure and disease)
o Ecological (population)
Correlational
Generate hypotheses
Unit: groups of people
Advantages: use populations with differing characteristics, and
existing data can be used
Test for potential confounders
o Cross-sectional (individual)
Measure of exposure at a given point of time
Key question: does the exposure precede or follow the effect?
For fixed exposures (ethnicity, blood type)
Advantages: the timing of the exposure is known (whether it
happened before or after the exposure), and looking at the
possibility causes of disease can make the study successful
If exposure is not measured before the disease the study gives
disease prevalence
Recall bias is high (people might not remember how much they
smoke for example)
Trends can be measured if people are examined in several
points in time
o Case control
On people that have the disease
Backtrack what leads to the disease
Compared to people without the disease to see normal
occurrence with similar characteristics
Disadvantage: can be biased as the past is examined, and it is
hard to find control groups that are very similar
Relative risk is calculated (compared to controls)
o Nested case control studies
Using cases from cohorts for which exposure information is
already available (checking whether they had the disease or
not)
Cohort is already designed
o Cohort
Aka follow-up study
Start with healthy people (see whether they get the disease
during the follow-up)
Can be pro and retrospective study
(Difference from case control is that the subjects are healthy to
start out with)
Advantages: insight into pattern of exposure and the disease,
and insight into causality
, Disadvantages: expensive, and large group is required
Experimental study (assign exposures, manipulate certain factors and measure the
effect)
- Randomized control trial
o Randomly selected people from two groups and are compared
o E.g. vaccine studies
- Field trial
o On healthy, but presumed to be at risk
o Data collection takes place in the field
o E.g. lead levels in children for those that were raised under lead
regulated environments compared to those that were not
- Community trials
o Do not focus on the individual
o Good to measure disease that affect the community (eg
cardiovascular disease)
o See the effects of social conditions on disease development
Sources of error:
- Random error (eg variation in people and by chance in small sample)
- Systematic error
o Selection bias
o Measurement error
Recall bias
Observer bias
o Response bias
- Confounding
