Cohort Studies

Sampling is based on

Exposure

Cohort study design

• exposed vs not exposed

• looking at those in each group that disease develops vs not

Cohort study (general)

• determines the association between exposure and disease/outcome

• starts with exposure status and moves forward in time

• outcome is not present in the cohort at the time the exposure occurs

• describes the incidence or natural history of a disease/condition in a cohort of individuals

Prospective Cohort Study

• begin in present in progress forward, collecting data from subjects whose outcome lies in the future

• data is non-existent at the time the study starts

• info about the outcome is collected in the future

Retrospective Cohort Study

• begins and ends in the present but involves a backward look to collect information about events that occurred in the past

• data exists at the time the study starts

• all of the relevant events (exposures and outcomes) have occurred when the study is started but direction of the inquiry is still forward

Difference between Prospective and Retrospective Cohort

• major: whether all of the data is in existence at time the study starts

• prospective:

  • exposure and outcome are ascertained as they occur during the study

  • the groups are followed up for specified time, data is collected prospectively and incidence is measured

  • data nOT in ecistence at the time the study starts

• restrospective:

  • exposure/non-exposure is ascertained from past records and outcome is determined by collecting data forward in time in the records

  • this allows for quicker data collection

  • data exists at the time the study starts

How are case-control studies different from retrospective cohort

• case control: distinguished from cohort study based on the study beginning with diseased/non-diseased

• retrospective cohort: begins and ends in the present but involves a backward look to collect info about events that occurred in the past

When is a Cohort Study Warranted?

• when good evidence suggests an association of a disease (outcome) with a certain exposure or exposures

  • evidence from clinical observations or case-control or other types of studies

• not ideal for rare diseases/outcomes

  • would take a long time for follow up

Design Issues

• clear definition of criteria for exposure and outcome

• start with the population

• exposed group:

  • where does the cohort come from? How were they selected?

  • could this influence the outcome (internal validity)

  • could it influence generalizability (external validity)

• non-exposed:

  • should be as similar as possible to the exposed group with respect to all other factors (confounders) that may be related to the outcome

Biases in Cohort Studies

• selection bias

• information bias

• surveillance bias

• bias from loss to follow up

Information bias

difference in info quality from exposed vs non-exposed

Surveillance bias

one group followed closer and outcome observed more frequently

Bias from loss to follow up

• results may be difference in those who had longer follow up

  • more likely to develop the condition?

• concerning if loss to follow up is >20% of cohort

Advantages to Cohort Studies

• efficient

  • restrospective: all relevant events already occurred

  • prospective: don’t have to rely on past data, recall bias not a problem

• calculate incidence of outcome (number of new cases over time)

• ideal when interval between exposure-outcome is short

• ideal when interval between exposure-outcome is short

• establishes temporal exposures and outcomes

• prospective- more complete data collection as investigators can dictate what data is being collected

Disadvantages to Cohort Study

• time consuming (prospective)

• concerns with data completeness, accuracy, quality (retrospective)

• follow up time may be too short

• confounding: known and unknown (can’t control for something if you don’t know it exists)

Relative Risk or Risk ratio

(Risk in exposed)/(Risk in non-exposed)

RR < 1

• risk in exposed < risk in non-exposed

• negative association; sometimes interpreted to be “protective”

RR = 1

• Risk in exposed = risk in non-exposed

• no association (value if the null hypothesis is true)

RR>1

• risk in exposed > risk in non-exposed

• positive association