Lecture 2B - 4 Study designs

Question 1

Define epidemiology

Answer 1

running a study (sample) that allows us to infer the population

Descriptive or analytical epidemiology does not ever achieve an exact match of the population

Question 2

what are the 4 main study designs you need to know ?

Answer 2

Descriptive studies -

Ecological study
Cross sectional study

Analytical studies -

Case control study
Cohort study

Question 3

Explain Ecological study

Answer 3

Done in the present, and we count the cases by groups of the population

ie number of people with aids in Liverpool, or males aged 50-60

• Identify groups to study
• Define what needs to be studied
– exposure and – outcome
we can :
– counting categorical data (nominal or ordinal)
or
– measuring continuous data (interval or ratio scale)
• Gather data on the group-level characteristics

Question 4

what are the issues with an ecological study ?

Answer 4

specific issues,
• Definition of characteristics

• Measurement variation

• Confounding
– ecological fallacy: falsely inferring individual-level
association from group-level association

• Chance (Random error)

Question 5

Explain Cross-Sectional studies

Answer 5

We find cases as a portion of the whole population
we are looking for prevalence

Random sampling is the best method to sample

we have to generalise study to the whole population

Question 6

what are the issues with a cross-sectional study ?

Answer 6

• Chance (Random error)

Most Important
• Sampling bias
• Responder / Participant bias

Question 7

Explain a case control study

Answer 7

always looking back on past events
what was a persons level of exposure in the past

you find cases in the pop, and you define controls within the pop

ask the same questions about levels of exposure

we compare the two groups based on possible casual factors

Question 8

what are the issues with a case control study ?

Answer 8

In addition to previous issues,
• Selection bias
– controls should reflect study population, and
– controls should be comparable to cases
• Information bias, especially for exposure
– differential misclassification
– non-differential misclassification
• Confounding
and as always:
• Chance (Random error)

we often have to adjust for other confounding factors

Question 9

explain a cohort study

Answer 9

Find exposed and unexposed groups within the population

give it time to the future outcome of the two groups
compare the outcomes of the two groups

we are comparing the incidence rates for the two groups

Question 10

what re the issues with a cohort study ?

Answer 10

Loss to follow-up
– differential loss
– survivor bias - only those who dont get sick will continue study
• Information bias, especially for outcome
– differential misclassification
– non-differential misclassification
• Confounding
and as always:
• Chance (Random error)

Question 11

questions are likely to ask - what type of study is this ?

when describing a study PICO is a good checklist, what is PICO

Answer 11

What is the study design?
P What is the Population to be studied?
I What is the Intervention / Exposure of interest?
C What is the Comparison / Control of interest?
O What is the Outcome of interest?

Question 12

what are the analysis techniques for the different study designs

Answer 12

• Descriptive epidemiology study designs:
– ecological study: unit of analysis is groups
– cross-sectional survey: unit of analysis is individuals
• Analytical epidemiology study designs:
– case-control study: analysis only odds ratio
– cohort study: analysis can be rate or odds ratio

Question 13

give some other comparisons of the study on generic issues

you do not need to remember all of this

Answer 13

Cost and time needed - ecological is cheap/low, Cross-sec and case-control are medium, cohort study is expensive/long

all of them suffer random error and confounding and a sampling bias

cross-sectional, case control and cohort have a recall bias

case control and cohort have a follow up loss and selection bias