Chapter 3: Cohort Studies And Case Control Studies Flashcards
What is a cohort study?
Observational study design where the starting point is exposure to a particular risk factor
Steps to a cohort study
- Determine exposed or not exposed
- This forms the cohort group
- Determine outcome
What is the direction of inquiry in cohort studies
Direction of inquiry is forward in time
Types of cohort studies
- Prospective (current)
- Retrospective (historical)
- Ambidirectional
Advantages of cohort studies
- Exposure is measured at the start prior to outcome – measurement of exposure of not biased by the presence of absence of an outcome
- Temporal sequence of events can be established
- Can provide information about time course of outcomes, natural history
- Can examine multiple outcomes
- Rare exposures can be study (not necessarily rare outcomes as you will need a large cohort)
- Can measure incidence rates directly
Disadvantages of cohort studies
- Slow and usually expensive
- Inefficient for rare diseases
- Differential loss to follow-up can introduce bias
- Exposure status can change
- Can become historical
How to conduct cohort study
- Research question
- Identify study population
- Measure exposure
- Follow up participants and determine outcomes
- Data analysis
- Interpret results
- Assess possible sources of error
Interpreting results in a cohort study
If RR = 1 - no association
If RR > 1 - risk in exposed greater than risk in not exposed
If RR < 1 - risk in exposed less than risk in not exposed
What is risk ratio and when to use it?
- When follow up time is identical (or nearly) for all cohort members
- For both the risk ratio and the risk difference it would be best practice to calculate a 95% confidence interval (CI). Statistical construct that provides information about a range in which the true values of the risk estimate lies within a 95% probability
Formula for risk ratio
Risk in exposed/ risk in not exposed
** to calculate risk - outcome(yes)/ total
What is rate ratio and when to use it?
- When there is variation between cohort members in duration of follow-up time
- Person time can be measured as the total number of days, months, years of follow up. Unit will depend on time between exposure and outcome
- Best practice to calculate corresponding 95% CI ratio
Formula for rate ratio
Rate ratio = rate in exposed/ rate in not exposed
*rate in exposed = outcome(yes)/ person-time in exposed
*rate in non-exposed = outcome(yes)/ person-time in not exposed
What is personae time?
- Particularly important in a cohort study that is dynamic
- For an individual: accounts fro the total time period during which an individual could develop/ is at risk of the outcome of interest
- For a population/ cohort study: total person-time risk is the sum of all individuals person-time, the total time during which outcomes could occur and would be considered events in the population of interest
What is a case control study?
Case control study is a type of observational study in which two existing groups differing in outcome are identified and compared on the basis of some supposed causal attribute
Advantages of case control study
- Relatively quick and inexpensive, can have relatively small sample sizes
- Useful for studying rare outcomes
- Useful for studying diseases with long latent periods
- Can study multiple exposures
Disadvantages of case-control studies
- Selection bias
- Information bias
- Possibility for reverse causality
- Not suitable for studying rare exposures (unless choose nests c-c study)
- Cannot estimate disease incidence or prevalence
How to conduct a case control study
- Research question and population
- Selecting cases
- Selecting controls
- Matching appropriately
- Matching exposure
- Analysis of case control studies
- Assess possible sources of error
Analysis of case control studies
- Compares frequency of exposure among the cases with the frequency of exposure among the controls
- Cannot estimate incidence from a case-control study
- Compare the odds of exposure among cases and controls suing the odds ratio(OR)
Formula for odds ratio
Odd of exposure (cases)/ odds of exposure (controls)
*odds of exposure (cases) = exposed cases/ not exposed cases
*odds of exposure (controls) = exposed controls/ not exposed controls
Assessing possible sources of error in case control studies
- Chance, bias or confounding
- Selection bias in selecting cases and controls
- Information bias in obtaining exposure information from cases and controls (recall bias a particular concern in self-reporting)
- Misclassification bias (how exposure or outcome is identified) – is it non-random (non-differential)?
o When proportion misclassified in the same in cases and controls = non-differential e.g. medical records
o Non-differential misclassification differs between cases and controls; this means that the association will either be under-estimated or over-estimated - Reverse causality is when the exposure is a consequence of the outcome
o E.g. poor infant growth lead to increased breast feeding, breast feeding did not lead to poor infant growth
o To limit this, exposure time period of interest should be limited
What are the two types of epidemiological study design?
- Interventional studies
o Clinical trials
o Experimental study designs - Observational studies
o Descriptive studies: ecological studies, migrant studies, cross-sectional studies
o Analytical studies: cross-sectional studies, case-control studies, cohort studies